Document [original]

IT-supported

Visualization and Evaluation of

Virtual Knowledge Communities

vorgelegt von

Diplom-Kaufmann

Matthias Trier

aus Berlin

Von der Fakultät IV – Elektrotechnik und Informatik

zur Erlangung des akademischen Grades

Doktor der Ingenieurwissenschaften

- Dr.-Ing. -

genehmigte Dissertation

Promotionsausschuss:

Vorsitzender: Prof. Dr. Sahin Albayrak

Berichter: Prof. Dr. Hermann Krallmann

Berichter: Prof. Dr. Norbert Gronau

Tag der wissenschaftlichen Aussprache: 04.04.2005

Berlin 2005

D83

Visualization of virtual Knowledge Communities i

ACKNOWLEDGEMENT

‚Mankind are to be taken in groups, as they have always

subsisted. The history of the individual is but a detail of

the sentiments and thoughts he has entertained in the

view of his species: and every experiment relative to this

subject should be made with entire societies, not with

single men’ (Ferguson, 1767).

This book reflects my past five years of work in the field of Knowledge Manage-

ment and Communities of Practice. During this time I learned much about my per-

sonal motives and values. I enjoyed the continuous challenges coming from daily

novel situations and the work with clever students that this job offers. I am happy

to have developed towards a very interesting and motivating topic with interna-

tional collaborations, which allowed me to meet people from all over the world,

who shared and fueled my interest and dedication to my research topic. However,

this work would not have been possible, if I would not have been supported by a

variety of people. First of all, I want to thank Prof. Hermann Krallmann for being

a constructive primary advisor, supporter and mentor in my professional career

and to Prof. Norbert Gronau for being a very approachable second advisor.

Further thanks go to Prof. Steven Seidman for inviting me to work at his faculty at

NJIT in New Jersey and for being a pioneer in the field I am working. He has

given me orientation and spurred my interest in my research. Prof. Marilyn Tre-

maine has given me personal motivation almost as an advisor and was a welcom-

ing chair during my visiting research stay in the US. She introduced me to many

like-minded people and let me participate in the Anglo-American way of doing

research which I am now able to compare to our German approach. Not to men-

tion the various wonderful talks with her about life and work. I thank Prof. He-

Kyung Cho, with whom I participated in an interesting research collaboration and

practical application of the developed prototype that literally spun the globe from

Korea over Germany to the US. Further, I want to thank a further faculty member

of NJIT, Prof. Roxanne Hiltz, for her constructive talk, orientation and motivation

and for simply being a dedicated researcher, who virtually started the discipline of

virtual learning and collaboration decades ago.

This work further also lives from the interesting perspectives and contributions of

practicing Knowledge Managers like Steven Tedjamulla. His practitioner’s view

of a Knowledge Manager in a global network software company continuously

challenges the approaches described in this book and generates new insights. It

ii M.Trier

provided me with insightful experiences in actually conducting a virtual collabora-

tion and group interaction across the globe.

Further thanks go to Michael Herzog, who I regard as a fellow colleague in two

academic institutions. He continuously provided me with inspiration, support and

motivation. I admire your organizational talent. Further I thank Claudia Mueller

for a wonderful year of joint work on our sophisticated e-learning offering for a

virtual university, for sharing my set of values and opinions as far as research is

concerned, and for helping me maintain speed and quality by numerous joint re-

views of papers. Thanks go to the team, consisting of Tilmann Bartels and An-

dreas Hoffmann for having contributed large parts of development work and for

sharing the inspiration and dedication to the prototype. Further I want to mention

Mathias Werlitz, Daniel Mueller, Marcus Walla and Daniel Graef for their techni-

cally outstanding development contribution and our resulting experiences in group

and project based software engineering management.

I thank my colleague Dr. Marten Schoenherr for introducing me to the general

topic and for awakening my interest, further Stephan Aier, Marian Scherz, Serkan

Tavasli and the other fellows for moral support and being a great pack of col-

leagues. I also want to mention ex-colleague Boris Wyssusek who has been an in-

visible source of inspiration and affirmation. I enjoy being able to have a look at

his deep and rich analysis of underlying organizational theories and I admire his

dedicated conviction for emphasizing the importance of abstract thinking, phi-

losophy and methodology in an engineering discipline - against all barriers. It

gave me orientation with the major challenge of moving from almost nihilist so-

ciological and philosophical concepts towards a tool which for sure (again!) leaves

out most of the vagueness, complexity and also important interrelationships but

still tries to technically interpret, cover and support the biggest possible scope.

To CJ and Perry I send my regards over the ocean. You gave me a great time dur-

ing my research exchange in New York and New Jersey and enrich my life.

Thanks go to my parents for their dedicated social and moral support and their un-

derstanding my temporary mental and physical absences. Further to Ulrike for

providing me with strong personal support and for simply being a warm sunshine

– even in days of winter.

And thanks to you friend or colleague, whom I have not mentioned but also have

not forgotten, for having had interesting talks about the topic, for showing, that the

interest in the topic remains and grows, for taking opposite and challenging per-

spectives or simply for sharing times of complete recreational distraction of the

sometimes overwhelming workload.

I learned much, from all of you!

Matthias Trier, March 2005

Visualization of virtual Knowledge Communities iii

CONTENTS

ACKNOWLEDGEMENT....................................................................................... i

CONTENTS ..........................................................................................................iii

OVERVIEW AND OBJECTIVE .......................................................................... vi

1 The Economic Development and the Event of Knowledge Work..1

2 The Development of KM as a Research Discipline .......................10

2.1 Towards a Working Definition for the Term Knowledge..................... 11

2.1.1 The philosophic Roots .................................................................. 11

2.1.2 Recent Reception of the Philosophical Approaches...................... 13

2.1.3 Examining Boundaries and Properties of the Term Knowledge... 14

2.2 Reviewing existing KM Approaches .................................................... 22

2.2.1 Preliminary Concepts related to modern Approaches of KM ....... 22

2.2.2 The first Wave of KM Concepts – Push Orientation .................... 25

2.2.3 The second wave of KM concepts – Pull-orientation ................... 39

2.2.3.1 Process oriented KM Approaches............................................ 42

2.2.3.2 Community oriented KM Approaches ..................................... 45

2.2.4 Current Reception of KM in Corporate Implementations............. 48

2.3 Resulting Framework of theoretical KM Approaches .......................... 50

3 A network oriented Foundation of Knowledge Management in

Organizations - Knowledge Networks................................................... 53

3.1 Early Organizational Theories and their Role for KM.......................... 54

3.2 Shift towards Organizational Sociology ............................................... 58

3.3 Systems Sciences in Organizational Theories....................................... 61

3.4 Networks, Coordination and recent Organizational Theories............... 68

3.5 Organizational Network Structures....................................................... 73

3.6 Implications for organization and network oriented KM...................... 75

4 CoP as an Instance of a Network Organization and an Instrument

of network oriented KM.........................................................................79

4.1 Definitions and Basic Properties of Communities of Practice.............. 80

4.2 Typologies of Communities.................................................................. 85

4.3 Community of Practice vs. other organizational forms......................... 87

iv M.Trier

5 Organization and Coordination in CoPs........................................91

5.1 Integrating Communities into existing Organizational Structures........ 91

5.2 Practical Examples for CoPs in Organizations ..................................... 93

5.3 Benefits for Organizations .................................................................... 97

5.3.1 Short term Benefits of Communities of Practice........................... 97

5.3.2 Social Capital................................................................................ 98

5.3.3 Trust............................................................................................ 101

5.3.4 Understanding Costs to evaluate Benefits................................... 102

5.4 Structural Properties of CoPs.............................................................. 102

5.5 Roles in a Community of Practice ...................................................... 105

5.6 Management of Community Structures - Pro and Cons..................... 107

5.7 What and How to manage................................................................... 111

5.8 Tasks and Processes during the Community Lifecycle....................... 114

5.8.1 Wenger’s Community Lifecycle................................................. 115

5.8.2 IBM Lifecycle............................................................................. 116

5.8.3 Integrated Lifecycle .................................................................... 117

5.8.3.1 Preparation Stage.................................................................... 118

5.8.3.2 Build-time Stage..................................................................... 119

5.8.3.3 Run-time Stage....................................................................... 120

5.8.3.4 Decline Stage ......................................................................... 122

5.9 Role of Measurement for Managerial Community Tasks................... 124

6 Information Technology to support Communities...................... 127

6.1 The Role of IT for running Communities ........................................... 127

6.2 IT Applications for Communities ....................................................... 129

6.3 Computer-mediated Communication in Virtual Communities ........... 137

6.4 Focusing Discussion Groups............................................................... 142

6.5 Current Gaps and required Concepts .................................................. 145

7 Towards CoP Transparency, Measurement and Evaluation.....147

7.1 Measurement and Success Factors for Knowledge Networks ............ 147

7.1.1 Measures from logging Community Activity ............................. 153

7.1.2 Measures from Group Theory Research ..................................... 156

7.1.3 Measures from Social Network Analysis.................................... 158

7.1.4 Measures for Social Capital or Trust........................................... 162

7.1.5 Measures for Knowledge Processes............................................ 164

7.2 Deriving a Measurement Concept for Communities of Practice......... 164

7.3 Social Translucence - adding Visualization to Measurement ............. 170

Visualization of virtual Knowledge Communities v

8 A Tool for visualizing, analyzing and modeling Communication

Networks of Knowledge Communities................................................ 174

8.1 Current Visualization Approaches...................................................... 175

8.2 The Tool’s underlying Graphical Visualization Method .................... 182

8.3 Software Engineering Process and Software Framework................... 184

8.3.1 Requirements Profile................................................................... 185

8.3.2 Software Architecture................................................................. 187

8.3.3 Data Sources and Data Architecture ........................................... 189

8.3.4 Intelligent Backend and Connectors ........................................... 194

8.3.5 Frontend and User Interface........................................................ 198

8.4 Applied Technical Frameworks and Algorithms ................................ 202

8.4.1 Graph Theory and Information Visualization............................. 202

8.4.1.1 Social Network Visualization and Analysis........................... 206

8.4.1.2 Spring Embedder Algorithm.................................................. 208

8.4.1.3 The application of Java 3D..................................................... 210

8.4.2 Animated Longitudinal Visualization and Analysis ................... 213

8.4.3 Semi-manual Content Coding and Analysis ............................... 215

8.4.4 Automated Shallow Text Analysis.............................................. 216

8.4.5 Implementing and extending Measurements............................... 218

9 Case Studies for IT-supported Community Visualization and

Analysis .................................................................................................. 221

9.1 Case Public Slashdot Discussion Group ............................................. 221

9.2 Case Instant Messaging in a Manufacturing Company....................... 223

9.3 Case Public Java Developer Forum .................................................... 229

9.4 Conclusion .......................................................................................... 232

10 Outlook: Towards IT-supported People Network Management235

11 Literature........................................................................................238

12 Appendices......................................................................................266

12.1 CoPs in corporate KM Approaches – Overview................................. 266

vi M.Trier

OVERVIEW AND OBJECTIVE

According to an analyst prediction, by 2007 individual’s time spent interacting

with others in the virtual world will exceed physical interactions by a factor of 10

to 1. This impressive example shows that electronic media are becoming one of

the main means for interaction. New technical solutions emerge and change the

availability of communication, but also its speed and range. Especially, the appli-

cation of these new communication channels for Knowledge Work in complex

work domains is a development with much potential. Here, the mode of experts to

form networks of contacts, from which they can draw resources to solve their

business issues, is of increasing relevance for the competitiveness of the company.

The related discipline of Knowledge Management has to react to this trend and

needs to establish instruments for a systematical support of such knowledge and

expert networks. The most relevant concept is called Community of Practice

(CoP) or, if it utilizes electronic media as its primary means of communication,

Virtual Community. The establishment of instruments for such network oriented

Knowledge Management directs the attention to the role of the discipline of Busi-

ness Informatics. With its research focus on developing innovative software appli-

cations to realize untapped corporate potential and thus to increase a company’s

competitiveness is the right perspective for the establishment of a novel IT support

for Knowledge Management in Virtual Communities.

To develop the according approach together with a supporting software solution,

this work has to take on a major challenge: As expert networks inherently build on

social mechanisms, this book needs to bridge the wide gap between the necessary

sociological analysis (which often enough is not guiding towards solving a prob-

lem) and the constructive and technical engineering of a concrete software system.

This leads to the following sections:

After analyzing current economic indicators to prove the necessity of Knowledge

Management for maintaining competitiveness (chapter 1), the basic foundations of

the discipline are analyzed (chapter 2). This includes a detailed discussion of the

core term ‘knowledge’ itself (part 2.1) as well as a comprehensive overview of the

development of KM approaches (part 2.3). This results in a timeline which shows

the development of the research discipline of KM during the past decades.

In chapter 3, the underlying complex theories of systems sciences and sociology

are developed towards an overview about properties and requirements of modern

and complex network organizations. As a result, in part 3.6., novel and concrete

implications for a modern and network oriented approach to KM are derived from

this discussion.

Visualization of virtual Knowledge Communities vii

In chapter 4 Communities of Practice are identified and described as a major re-

cent concept, which is an actual instantiation of a networked organization for or-

ganizing Knowledge Work in expert groups. Chapter 5 specializes on describing

the main properties, roles, and processes of a community and its development

through lifecycle stages.

The resulting picture of the basic mechanisms is then extended in chapter 6 with

an extensive discussion of Information Technology to support the expert net-

works. However, the analysis results in the insight, that current IT is not at all sat-

isfying the requirements of virtual Knowledge Communities in corporate applica-

tions. Especially, the important role of the community moderator and manager is

unsatisfactorily supported. This person needs transparency about the large group

he is responsible for. This implies the necessity of instruments for monitoring,

measurement and evaluation, which is also emphasized by thought leaders and

major institutions in the CoP area. Further, the sociability of the expert group

needs to be improved. To address these issues, chapter 7 develops a comprehen-

sive measurement system for analyzing virtual Knowledge Communities. It con-

sists of 55 factors in four domains and draws its measures primarily from socio-

logical domains, such as Social Capital and Trust research and Social Network

Analysis; but it also includes Knowledge Processes and plain structural analysis.

To implement the conceptualized support for CoPs with appropriate measures and

visualizations, an extensive software solution which aids as an add-on to current

community platforms has been developed. It is described in chapter 8. The pri-

mary challenge was to create insightful visualizations, which integrate 2D and 3D

Graph Drawing Techniques for Social Network Analysis with Topic and Keyword

Analysis methods and to merge this conglomerate with the measurement system.

Finally in chapter 9, three case studies are introduced to illustrate the application

of the software solution and its benefits for providing a CoP moderator or manager

with detailed insights about the structure and processes of his group.

In summary, this work addresses the following main questions:

1. What is the role of Communities of Practice in Knowledge Management?

2. What are necessary conditions for Knowledge Management concepts in

order to be able to support Knowledge Work in People Networks?

3. How can communication networks of Virtual Communities be modeled

in order to be analyzed and visualized to support moderators, analysts,

and members?

4. What data offered by available communication means of CoP software

provides most value for visualizing, analyzing, evaluating, and develop-

ing the virtual knowledge community?

5. How can Social Network Visualization be synergistically integrated with

Topic and Keyword Analysis?

6. How should a software solution for automatic analysis of virtual expert

groups and subsequent management support be designed?

Visualization of virtual Knowledge Communities 1

1 The Economic Development and the Event of

Knowledge Work

Understanding the objective of the discipline of Knowledge Management requires

examining the event of Knowledge Work. Being largely based on expertise, this

special type of work slowly grew in importance over the last decades and was fu-

eled by recent industrial developments, which include:

• shortened Product Life Cycles, despite

• increased product development efforts, subsequently

• increased product and process innovation and

• increased investments in Research and Development,

• increased market share of high-tech industry,

• increased levels of education, which allowed for

• increased complexity of offered (and customized) solutions, which

• intensified customer relationships, and lead to

• the augmentation of core offerings with (Knowledge intensive) Services.

This chapter aims at substantiating these underlying economic trends in order to

arrive at a very concrete representation and a clear understanding of the current

industrial evolvement towards the necessity of managing knowledge in enter-

prises. This situational review will provide the foundation for the subsequent theo-

retical discussion of the simultaneously emerging research field of Knowledge

Management, which continuously strives to offer methodological approaches and

instruments which meet the requirements of current economic needs.

Analyzing recent industrial developments, it can be recognized, that enterprises

are increasingly facing the problem to manage highly innovative, complex and re-

search-intensive products in mature markets. They have to survive in a well estab-

lished competition with low price margins. This can be illustrated by analyzing

the profit margin (i.e. return on sales) as an approximation. For example, in the

German machinery sector, this margin fell from 3.5 percent in 1997 to a more or

less stable yet small 2 percent in 2003 (Dresdner Bank, 2002).

Another challenging development is the dramatic shortening of Product Life

Cycles (PLC). In the machinery industry it decreased from twelve years in the

seventies to about seven years in the nineties. In the Information Technology sec-

2 M.Trier

tor, it even approached a record-low 5.3 years, which is less than half the duration

of 11.1 years during the 1970s (Droege et al., 1993).

Although Product Life Cycles decreased, product development times increased

from an average of 1.6 years in 1978 to 4.2 years in 1994, thus leading to a con-

flict situation for businesses (Perillieux, 1991:26): Higher investments have to pay

back in a shorter period of time.

These market pressures where amplified during the recent major period of reces-

sion. In the European Union, the growth of the Gross Domestic Product1 (GDP)

went from 3.5 percent in 2000 back to just 0.6 percent (declining 2.9%) and was

accompanied by declines of 3.4 percent in Japan and 0.2 percent in the United

States (OECD, 2002b). A final source for competition was provided by the intro-

duction of the single market. Since that, it contributed to a 30 percent increase in

intra-EU trade of manufactured goods since 1992. Further it boosted direct in-

vestment within the EU. In 2000, these flows were twelve times greater than in

1992 (OECD, 2002a).

Next to the single market, the decreasing revenues in established locations pushed

enterprises towards more globalization. Globalization can be indicated by meas-

uring the amount of foreign direct investment. It has grown impressive 15 percent

per year from 220 billion dollar in 1989 to 768 billion dollar in 1999 (OECD,

2000). Corporate activities mainly were aiming at „lowering production costs

and/or gaining access to new markets. Firms engaged in labor intensive manufac-

turing, for example, sought to decrease costs by locating labor intensive operations

in countries where wages were substantially lower“ (OECD, 2002a:15).

Next to the foreign investment, strategic alliances show, that business is becoming

more internationally oriented, networked, and complex. Between 1990 and 1998,

5100 strategic alliances were formed between enterprises (UNIDO, 2002:15).

Even closer cooperation can be achieved via Mergers and Acquisitions (M&A).

They increased by more than 500 percent between 1990 and 1999, rising from 153

billion to 792 billion dollars (OECD, 2001). These alliances allowed the access to

new markets without investing in local production structures and enable the com-

panies to concentrate on perfecting their profitable core competences (e.g. using a

cost or a quality strategy), while simultaneously engaging in a complex coopera-

tion network.

The short period for investments to pay back has not only been met by cost reduc-

tions, inter-organizational networks, and market entries but also by increased cor-

porate investments in rather complex Research and Development (R&D) en-

1 GDP is the total value of goods and services produced by a nation over a given period,

usually 1 year.

Visualization of virtual Knowledge Communities 3

deavors. For instance, in OECD countries2, enterprises accounted for 69 percent of

all R&D funding in these countries in the year of 1997 (UNIDO, 2002:15). There

is not only a large share of funds invested by enterprises, but also a strong growth:

Together, the four most important areas EU, US, and Japan raised their spending

from 255 billion Euro in 1995 to 453 billion Euro in 2001 (cf. Figure 1). This is a

growth of nearly 80 percent of research effort within just 6 years (European

Commission, 2003a:29).

Figure 1: Evolution of Business Expenditure on R&D.

Source: European Commission (2003a:29)

This strong growth in R&D efforts also translates into an increase of research

personnel: The business sector in the US, the EU and in Japan hosts about two

million researchers. In the EU, there is currently an average of 5.68 researchers

per 1000 employees. This number grew by 2.6 percent from 1996 (European

Commission, 2003a:44). Following that, the EU patent applications rose by 10

percent to 107 patents per million people. Figure 2 illustrates that the main growth

came from the Software and IT Services sector with 30 percent additional R&D

expenditure annually (volume in 2002: approx. 15 billion Euros). Among the

other contributing sectors were Pharma & Biotech (approximately 16 percent an-

nual growth to 35 billion Euro), the Automotive sector (10 percent to 37 billion

Euro), Engineering and Machinery (12 percent to 6 billion Euro) and Electronics

and Electrics Sector (8 percent to 22 billion Euro). It should also be noted that the

IT hardware sector hardly grew in R&D investments (only 1 percent) but remains

2 Twenty countries originally signed the Convention on the Organisation for Economic Co-

operation and Development on 14 December 1960. Since then a further ten countries

have become members of the Organisation. Among them are: USA,UK,G,F,I,E,S,J,A.

4 M.Trier

on the first position in terms of volume (2002: 45 billion Euro) (European Com-

mission, 2003a:33).

Figure 2: R&D Expenditure in selected Sectors – average annual Growth Rate, 1998-2002

and absolute level (in mill. EUR), 2002. Source: European Commission (2003a:33)

The aspect, that IT is the sector with the largest growth rate and with the largest

volume also relates to the event of the Information Society, with its substantial

cost savings and increases in speed of business processes. This can be illustrated

by looking at the number of internet hosts. In Europe, this number increased from

0.6 million in 1991 to 12.4 million in 1996 (OECD, 2002b). Equally, the number

of broadband connections doubled in the twelve months before October 2003 to

almost 20 million connections across the EU (European Commission, 2004). This

development is tremendously fueling the efficiency of business processes in inter-

organizational global networks.

The increasing importance of software to support business transactions can also

be substantiated by the growth in sold software products of 4.6 percent in the year

2004 in the European Union. According IT services grew by 2.6 percent in the

same year. The current approach of businesses to reap the benefits out of their ex-

isting structure can be substantiated by the decreased investment growth rate. It

fell from about 4 percent (period 1995-2000) to minus one percent in 2001 (Euro-

pean Commission, 2003a:12).

In such a competitive environment, two thirds of all companies frequently in-

vested in new innovations which they subsequently introduced into their organi-

Visualization of virtual Knowledge Communities 5

zation (ZEW, 2003). Taking the representative example of Germany in 1998,

about 70 percent of all money invested into innovations was allocated to product

innovations (IFO, 2001:74; also cf. Figure 3). The remaining 30 percent of inno-

vations were allocated to process improvements.

Figure 3: Targets for Innovation, Manufacturing Sector, Bavaria, Germany, 2000. Source:

(IFO, 2001).

These innovation investments are inextricably related to the objective of achieving

cost reductions. This can be illustrated by the following example. In 2001 in the

machinery sector, about 3 percent of costs could be reduced by the relatively sta-

ble share of process innovation investments. This is a much smaller reduction as

for instance in the automotive industry, which achieved 9 percent cheaper proc-

esses (ZEW, 2003).

In this situation of increased relevance of research and IT for supporting business

transactions, the market share of the high-tech industry started to grow. In 2001,

8.4 percent of the EU’s value added originated from high-tech and medium high-

tech industries. The respective growth rate of this sector’s value added was 1.84

percent for the EU between the years 1996 and 2001. The growth in the EU as a

whole was near 0.5 percent - substantially less than the growth in value added of

these industries for the same period. This is indicating important overall produc-

tivity gains in European high-tech manufacturing (European Commission,

2003a:78). Simultaneously, the share of medium- and high tech products in global

exports rose from 58 percent in 1980 to 65 percent in 1997 (UNIDO, 2002:15).

To cope with the dynamic markets due to short Product Life Cycles, increased

globalization, production complexity, and high-tech products, the employees re-

quire higher degrees in education and frequent training.

„High-tech employment in 2002 was 4.5% higher than in 1997 in the EU as opposed

to an increase of only 2% in total manufacturing, (European Commission, 2003c),

6 M.Trier

suggesting a strong link between R&D intensity, job creation and competitiveness.“

(European Commission, 2003a:80)

Following this trend, between 1998 and 2001 the EU produced 14 percent addi-

tional graduates. In the same area, in 2002 about 8.5 percent of the population

aged 25-64 have followed continuous education or training. This number grew by

7.9 percent compared to 1997. (European Commission, 2003a:50). Of course this

is influencing labor productivity (i.e. GDP per hour worked). It increased in the

EU countries from 1997 to 2002 annually by 1.48 percent (in the US by 2.13 per-

cent).

This efficiency effect of a qualified workforce can be illustrated by the drop of the

total number of employees in the UK automotive sector from approximately 290

thousands in 1998 to 240 thousands persons in 2001 - although the sales value re-

mained at a stable 42 billion Pounds. Gross Value Added as percentage of labor

costs fell from 169 percent in 1997 to 126 percent in 2000. Compared to the small

GDP change, this is indicating that with lower labor expenditure more Gross Do-

mestic Product has been generated. The increased automation of ‘simple’ transac-

tion processes, which reduce the demand for unskilled labor, is also observable in

this industry: Stock Turnover increased from 6.6 in 1995 to 9.7 in 2001 indicating

lower stock and faster and more efficient production processes (DTI Automotive

Unit, 2004).

Another impressive indicator can be found in the machinery industry of Germany.

Here, between 1980 and 1996, the overall amount of employees decreased by 13

percent. During that time, people without a certificate of their professional quali-

fication did reduce by 51 percent whereas the share of people with a university

degree rose by 64 percent. This shows the increasing average qualification levels

of employees as the machinery sector is behaving similar to other branches (Licht-

blau, 1998).

The increasing average qualification of employees implies the increasing role of

the utilization of intellectual (intangible) assets as a major incentive for manage-

ment:

„Increased attention is being paid, for example, to the importance of the capabilities of

human resources to firm performance – which would include the knowledge, skills

and abilities of systems engineers, programmers and researchers.“ (OECD, 2002a:15).

The skilled labor, the event of complex high-tech products, and sophisticated In-

formation Technology lead to the extension of core products by various augment-

ing services in order to cover the complete customer buying cycle including pre-

sales, sales, and after-sales services. New augmenting business units (e.g. decon-

struction, services) were emerging and often enough, they developed to become

the most profitable business segments and growth areas around a particular mature

technical solution. The general increase in services as compared to manufacturing

can be substantiated by a recent survey. In 1958, 23 percent of the population of

Visualization of virtual Knowledge Communities 7

Belgium, France, Germany, Italy, Luxembourg and the Netherlands worked in a

farming job, 40 percent of the population worked in industry. By 2001, farming

jobs have dropped to 4 percent and industrial jobs have dropped to 29 percent for

the then 15 EU countries. The opposite holds for the service sector: Its share grew

from only 37 percent in 1958 to 67 percent in 2001 and was hence the major

source for new jobs (European Commission, 2003b:5).

A special subset of services with a highly skilled workforce and the application of

expert knowledge is constituted of Knowledge intensive Services (KIS), which

tremendously increased in importance.

„Knowledge intensive services are defined according to the Eurostat definition as:

post and telecommunications, computer and related activities, research and develop-

ment, water transport, air and space transport, financial intermediation, real estate,

renting and business activities, education, health and social work and recreational,

cultural and sporting activities.“ (Eurostat, 2004)

The development of KIS is closely linked to the growing specialization of indus-

tries and the need for even more specialized services emanating from other service

and manufacturing sectors. In 2001, KIS accounted for nearly 39 percent of EU’s

services’ total value added. In Germany it even approached 50 percent. The value

added produced by KIS shows positive growth over the period 1996 to 2001. On

average the annual real growth rate has been 4.5 percent for the EU as a whole

(Germany: -0.31 percent, UK 17.6 percent). In 2001, 37.3 percent of people em-

ployed in the EU worked in KIS causing a growth rate of 3.23 percent between

1996 and 2001 (European Commission, 2003a:83).

Developing Knowledge intensive Services plus complex augmenting services

around the core product to cover the complete customer buying cycle implies, that

vendors employ experts and extended their participation to include the planning

and conception stages of customer projects. They thus develop towards project-

driven business and invest much effort and upfront investments into applying their

available expertise to the special customer demands and requirements. Due to this

extended relationship between customers and vendors, very qualified sellers and

technical experts can learn very much about the customers’ problems and can

more specifically apply their methodical and technological knowledge. This kind

of occupation is increasingly important compared to the conventional production

or the selling processes. The focus is moving towards a solution oriented business,

consisting of highly flexible planning in strategic networks, customization, inte-

gration and implementation of the offered complex bundles of services and prod-

ucts.

As shown in the statistics, in such an environment, the main value added is not to

be found in routinuous execution of business processes, which depreciates to be-

come a prerequisite (i.e. fast order processing using modern IT). Rather, the value

is generated with skilled experts and their knowledge intensive work (or Knowl-

edge Work) like technology consulting, configuration, or engineering. The flexible

8 M.Trier

management of exceptions in a complex customer project and the efficient execu-

tion of very individual customer requests become more important than physical

transactions (which are getting apt to automation). The according management

systems hence develop from optimizing transactions (like in business process en-

gineering and management) towards including and emphasizing the problem-

solving tasks needed for Knowledge Work in knowledge intensive activities.

Improving the current efficiency in such a knowledge-dependent business activi-

ties generally enables corporations to:

• improve time-related efficiency, due to increased speed of business proc-

ess execution,

• improve cost efficiency, due to reduced double work and more efficient

distribution of resources (including investments in knowledge genera-

tion),

• improve resource efficiency, due to reduced double work and shorter

process durations or due to better process designs,

• support the capability to cope with complexity, due to employing self-

regulating mechanisms,

• integrate important employees into the problem solving processes, by

more connected, more diverse, more current communication structures

between the people who execute knowledge intensive business processes,

• help to integrate related domains of the corporate value chain,

• help to integrate individual products and related services into a complex

solution,

• help to manage distributed locations (improving logistics by improving

knowledge sharing and information logistics, and

• improve the reflection about learning processes (eliminating errors) in a

company. This feedback can be incorporated in incentive systems and

process structures to reward positive structural ideas and changes for im-

proved problem solving. It further can

• establish clear paths for efficiently reacting on new information ‚sensed’

somewhere at the system’s interfaces to its customers, production, proc-

esses, suppliers, technologies, or employees,

• support the transfer of generated knowledge to save effort for solving

similar problems,

• support the transfer of knowledge between affiliations in countries with

different technological and economic status (e.g. IT in 1st, 2nd, and 3rd

world countries),

Visualization of virtual Knowledge Communities 9

• help to ‚package’ insights and knowledge as fixed structures to reuse

them as a general solution to a class of problems,

• strategically develop certain knowledge domains to maintain and im-

prove future competitiveness,

• identify and remove barriers for efficient structural adaptation as a reac-

tion to learning processes resulting from solving business problems, and

• support the transparent communication of corporate capabilities and po-

tent competencies to stakeholders in order to better substantiate ‚good

will’ and other tangible resources.

The list of benefits resulting from managing Knowledge Work and Knowledge is

long and their realization is eminent due to recent economic developments. But

how can this be achieved? The answer to this question is similarly the objective of

the research discipline Knowledge Management (KM), which will be the primary

perspective of this book. To prepare for an answer to this question and to identify

innovation potential for appropriate technical support, the next chapter discusses

the development of a body of theories around the idea of managing knowledge

and Knowledge Work by the means of Knowledge Management.

10 M.Trier

2 The Development of KM as a Research Discipline

The research field of Knowledge Management consists of various and only

loosely connected scientific concepts. These concepts can be segregated into two

developmental stages (or two ‘waves’) of KM. Prior to these two waves, there was

an initial stage, which discussed the role of knowledge and Knowledge Manage-

ment in an enterprise. This preliminary KM stage can eventually be traced back to

Penrose (1958) and Taylor (1911), which implies that knowledge has always been

an issue of interest for creating products and executing production processes (also

compare chapter 2.2.1).

Of course, a thorough discussion of the utilization of knowledge in enterprises ne-

cessitates a scientific analysis and a subsequent definition of the term knowledge

itself. Related approaches reach back thousands of years to Plato and his pupil Ar-

istotle, which created the first foundations for the western comprehension of the

term knowledge. However, it has to be added that no accepted definition of this

complex term with its multitude of aspects could be generated so far. For example,

only recently, the western scientific understanding of this term was challenged and

influenced by Nonaka and Takeuchi (1995) who criticized the western basic ap-

proaches (Cartesian Dualism) and emphasized Japanese approaches which again

added a new aspect to the topic.

Following these two main strands, the next sections will now first provide a short

introduction into the approaches that conceptualize the term knowledge, before the

preliminary KM stage and its development towards the discipline of KM is intro-

duced. Afterwards, relevant approaches of the first wave of KM theories are

briefly discussed and put into a framework which illustrates the applicability of

their insights. This reduction to various ideas and the comparative classification

shall enable to elicit and comprehend the diffuse theoretic body emerging in the

discipline of KM during this first wave.

Finally, the approaches and challenges of the current second wave of KM research

are added to identify the management of Knowledge Networks as a current chal-

lenge both for theory and practice, which subsequently constitutes the primary

perspective of this book. However, it is not intended to see it as a new discipline;

rather the roots and thoughts about this special view are collected and integrated in

this chapter to arrive at a transparent theoretic framework for the subsequent de-

velopment of practical KM solutions. All these theories are requiring a sound un-

derstanding of the core term ‘knowledge’ which will be discussed now.

Visualization of virtual Knowledge Communities 11

2.1 Towards a Working Definition for the Term Knowledge

On an abstract level, two scientific strategies are applicable to arrive at a concrete

definition of the term ‘knowledge’. The first approach is to determine the con-

cept’s boundaries and its relation to other concepts in order to indirectly encapsu-

late the term itself. The second way involves the development of various classifi-

cations of sub-types to reflect on the various special properties of ‘knowledge’

emerging from the differences between the sub-types.

Before these two strategies shall be applied to create a working definition of the

term knowledge and its related constructs, the theoretical discussion which

stretched over several thousand years will be briefly examined. It actually started

with first philosophical considerations which have created schools of thought that

developed a philosophical understanding of the term. This helps to provide a basis

for further explorations in the context of Knowledge Management.

2.1.1 The philosophic Roots

The philosophical debate about the meaning of the term knowledge is related to

the field of epistemology and started with Plato and his philosophical theory called

Rationalism. It assumes the existence of a priori existing knowledge. Plato calls

this the divine idea and sees the actually perceived world as its only partially and

fragmentarily perceived shadow. The only way to approach to the divine ideas and

concepts is by means of deduction using reasoning. Plato further adds the notion

of knowledge to be true: he sees knowledge (‘insight’) as a correct conceptualiza-

tion together with its explanation (Plato, 1977:18).

However, Plato’s pupil Aristotle disagreed with his teacher’s approach. His Em-

pirism assumes that no idea can exist without its sensual perceptibility - knowl-

edge is solely and inductively created by sensual perceptions. Following his ap-

proach, recognition results in memories, multiple memories result in experience,

and finally the stable and abstract essence of experiences results in knowledge

about a concept (Aristotle, 1993:83). Later, Wilhelm von Occam (Nonaka and Ta-

keuchi, 1997:65) tried to merge Platon’s and Aristotle’s perspectives by noting

that abstract knowledge and the associated deductive reasoning prerequisites pre-

viously perceived intuitive knowledge, derived inductively from existing ele-

ments. However, the two opposing concepts determined the subsequent discus-

sions. For example, Augustinus’ and later Descartes’ rational thinking followed

Plato (Descartes, 1979:108) whereas Thomas von Acquin and later also Locke

(1690:II§2) followed Aristotle. The opposition between Locke and Descartes even

developed towards the two powerful directions of British Empiricism versus

Continental Rationalism: Descartes wrote about his four strategies for rational

thinking whereas Locke wrote his renowned ‘Essay concerning Human Under-

standing’. There he concludes that only sensual perception can fill a mind, deduc-

tive reflection is the source which uses prior experience. The increasing impor-

12 M.Trier

tance of the notion of abstraction from experiences which began with the early

Christian philosophers has later also been identified by von Glasersfeld (1996:93).

In 1710, Vico explicates first constructivist arguments and partially preempts the

later works of Kant by stating, that reason creates knowledge via analyzing how

things are combinatorially constructed or produced (Vico, 1710).

A further integration was attempted by Kant, who assumed that knowledge is cre-

ated when sensual perception and logical thinking are combined (Kant, 1980:45).

However, Kant thinks of subjective and objective belief in something being true,

which shows that he somehow tends towards the rational approach, which was as-

suming something objective outside perceptibility.

Marx emphasized the importance of interactions with the (and influence on the)

environment resulting in an adaptation and thus adds the notion of action: Knowl-

edge is created in the process of acting with things in a situation (cf. Russell,

1954). Things are recognized in a process of activities focused on the thing. This

idea has later been further discussed by Heidegger (1986) and culminated in the

emergence of Existentialism: Knowledge is not only generated by a reflection

about perceptions, but by acting and reflecting on actions (also cf. Sartre, 1980)3.

Finally in 1907, the concept of Pragmatism emerged. Here, one of this school’s

philosophers, John Dewey, notes that knowledge is useless if it is not implemented

in actions to change the environment (cf. Nonaka and Takeuchi, 1997:40).

Later in that century, Constructivists like Ernst von Glasersfeld followed. They

based their approaches on the biological conception of knowledge discussed by

Piaget, who explained knowledge as a biological means of adaptation. This led

Piaget to forego the idea of knowledge as a representation of an external ontology

and led him towards an emphasis on cognition which results in structures for ad-

aptation (cf. von Glasersfeld, 1996:107). Extending this thought, constructivism

added the conception of a knowledge, which is not passively consumed, but ac-

tively created by cognitive functions for means of adaptation. Cognition is orga-

nizing experiences of the subject and is not able to create a final idea of an objec-

tive external ontology (von Glasersfeld, 1996:96). These more recent concepts

elevated the discussion of the term knowledge from philosophy to natural sci-

ences. The related disciplines which now increasingly contribute new insights

which impact the notion of the term knowledge are focusing on human cognition

(cognitive theory) and mental mechanisms to accumulate knowledge (connection-

ism, neural biology; cf. Anderson, 1995).

In summary, the philosophical discussion emphasizes the perception of the envi-

ronment to generate experiences (as the essence of multiple memories) and the

essence of experiences which comprise knowledge. The last aspect includes logi-

cal reasoning and rational thinking about practical experiences to further develop

3 This is obviously pre-empting the Learning Processes developed by Argyris and Schoen

(1996) which is described in chapter 3.3.

Visualization of virtual Knowledge Communities 13

them towards abstract principles. This rational thinking later has been augmented

with arguments for a ‘natural’ construction of a body of knowledge from past ex-

periences in order to adapt to the environment. Here, abstraction helps to improve

the insights stemming from initial experiences and thus to somehow approach an

abstract hypothetical ‘divine’ idea in a second step. All these aspects emphasize

the importance of researching human cognition. After all, only the application of

such mental processes in continuous interactions with the according objects of the

environment generates profound knowledge. The core concept so far hence con-

ceives knowledge both as a process and a result of a cycle consisting of action,

experience, cognition, reasoning, reflection, connecting and structuring informa-

tion, adaptation, and again action. It is left to neural biology to explain the actual

stored physical or informational essence which enables individuals to expand and

utilize their knowledge using their brains.

2.1.2 Recent Reception of the Philosophical Approaches

In order to arrive at a concrete picture of the current usage of term knowledge, it

has to be examined how the philosophical findings have been adopted by recent

working definitions in the context of Knowledge Management approaches. Here,

generally, it can be recognized that obviously for reasons of pragmatism - which

seems to be necessary to communicate the complex concept of knowledge - most

current KM approaches simplify the construct4.

The primary dichotomy between Rationalism and Empiricism as established in the

philosophical dialogue is being referred to again by Lehner (2000:143). In his dis-

cussion of KM approaches, he differentiates into knowledge as something objec-

tively given and knowledge as something constructed. Further he emphasizes the

sociological dimension of the term as it is always the knowledge of someone and

thus bound to (sometimes multiple) people5.

A rather detrimental approach, which restricts the added value of the concept of

knowledge, has been proposed by Daniel Bell: He defines scientific and theoretic

knowledge as a collection of structured facts and ideas, which allow a reasoned

judgment or an experimental result and can be submitted and transported by

means and media of communication (Bell, 1973:180). Although this conception

relates to experimental experiences (Empirism), the notion of communicability

restricts this definition to explicit knowledge (which can also be interpreted as in-

4 For some approaches it has even to be warned, that they not only simplified for pragmati-

cal reasons but arrived at a rather degenerated instrumentalization of the term knowl-

edge, mostly to sell a related, usually rather unreflected and often technological ap-

proach under the more sophisticated and sometimes even mystified brand ‘knowl-

edge’. These approaches are either briefly mentioned or even left out, as they did not

contribute (but rather hindered) the evolvement of the discipline of KM.

5 This aspect will be subject to a more detailed discussion in chapter 3.

14 M.Trier

formation). This may be useful in some scientific fields (i.e. engineering or com-

munication technology), but it is a quite dangerous starting point for Knowledge

Management.

The property of truthfulness is being put into the discussion again by Segler, who

sees knowledge as the matter, which actors use to generate behavior with different

degrees of ‘truthfulness’, reaching from objective insights to subjective beliefs

(Segler, 1985:187).

As a final example, the knowledge definition of the Association of German Engi-

neers (VDI) emphasizes the close relation to action: Knowledge is the structured

experience derived from possible, executed, or observed actions of the past includ-

ing the perceived objects and subjects (VDI, 1995). This notion of action is the

most important adoption in KM definitions of knowledge.

Next to the implementation of philosophical findings, KM concepts also try to cre-

ate an understanding of their core term and ‘object’ under consideration by identi-

fying borders or properties of knowledge. These approaches will now be dis-

cussed.

2.1.3 Examining Boundaries and Properties of the Term Knowledge

After introducing the generic philosophical ideas about the concept of knowledge,

two scientific strategies which help to advance the understanding of the term

knowledge can be employed. The first is to actually discuss differences to related

terms and hence indirectly and deductively define the boundaries of the concept to

better understand the concept itself. The second is to break the term down into dif-

ferent categories of knowledge with more comprehensible sub-types. The latter

strategy also leads to a set of descriptive properties to closer describe the qualita-

tive aspects of the knowledge of some issue.

The first and most important terminological boundary that needs to be explored is

between the terms knowledge and information. The most renowned definition of

the term information comes from Bateson (1972:315), who said that information is

a difference which makes a difference. To explain this short statement, Luhmann

adds that the system, which processes the information, isolates an information

from its environment, because it is relevant for the system as it changes the

knowledge of the system in an intended direction (Luhmann, 1984:68). Informa-

tion is actually only emerging when a system employs criteria for relevancies and

evaluates all the surrounding symbols and data with it (Willke, 1998:8)6. Watson

6 However, Willke creates the concept, that data gets information if it is put in a context of

relevancies. Later again (other) relevancies create knowledge from the information.

This system of double interpretation seems incorrect, because already the first upgrad-

ing procedure (data to information) requires the system to use internal criteria and

Visualization of virtual Knowledge Communities 15

(1996:25) supports this idea by defining knowledge as the ability to use informa-

tion purposefully. In the KM literature, Lehner (2000:142) gives a very illustrative

picture for this aspect: In an integrated cognitive system, information is the dy-

namical component and knowledge the static one. Information transports and

changes knowledge. The information is only information for that very system

(Lehner, 2000:142).

This subjective value of information directly relates to another important term:

meaning (or sense). Systems with knowledge need an objective for the informa-

tion to mean something for the system. Information can not be selected without

being relevant for the system. Otherwise the information which is provided to the

system is noise, without beneficial effects for the system (Luhmann, 1984:98).

This meaning is influenced by continuities and existing structures. Luhmann

states, that Information as differences also causes causality when it appears in cer-

tain patterns (like in a relation to a successful action, e.g. handling hot metal needs

gloves). On a neural level, this can be metaphorically illustrated by the activation

of higher neurons. They only react, if lower-level neurons have received a certain

input signal in a certain combination. Hence the simultaneous signals constitute

the causality for the higher neuron. The higher neuron can even be interpreted as

an indicator for having identified causality in the environment.

Luhmann puts the whole aspect of meaning into a short sentence: Meaning is the

difference between perceived actual situation and the intended and projected op-

portunities (Luhmann, 1984:100). This again leads directly to an intended success-

ful action (also cf. Marx and the Pragmatists in chapter 2.1.1) which again refers

back to recognition and information. It also relates to the idea of confidence in the

belief of something being true (cf. Kant, 1980, chapter 2.1.1) as this reflects the

confidence that the intended action or decision really results in the expected and

projected successful result.

This finally fuels the discussion of the value of having knowledge: The construc-

tion of a knowledge network consisting of past, combined, refined, evaluated, and

sometimes abstracted experiences (cf. Aristotle, 1993; chapter 2.1.1) yields in the

improvement of the quality of a system’s manipulation of its actions and decisions

(output) and eventually also leads to a better adaptation to its environment.

In a social system like an enterprise the environment necessarily contains other

people. In such a social system, an action is also interrelated with these persons.

Von Krogh and Roos (1996) build a relation between people networks and the

term knowledge. Knowledge is the result of interaction of people in networks7.

relevancies derived from prior experience (as what should provide an external set of

relevancies as opposed to the internal set for knowledge creation from information?).

7 North (1999) states that von Krogh/Roos define three perspectives on Knowledge:

Knowledge as the result of interaction of people in networks, knowledge as the result

of personal experience, and knowledge as the processing of information. However,

16 M.Trier

A last boundary needs to be drawn between the terms learning and knowledge. As

the gerundial form of the term learning implies, the emphasis lies on a process.

Weick (1991:122) notes, knowledge creation is a form of learning. Gueldenberg

(2001:159) gives a more detailed explanation: Learning is the processing and im-

plementing of perceived and selected information into existing knowledge. It

thereby also develops new knowledge. This is also supported by Lehner (2000).8

However, this notion does not actually include the aspect of learning from abstract

reasoning using the existing knowledge only and thereby restructuring and devel-

oping it further (or simply ‘thinking’).9

By now one gets the impression, that knowledge is an amorphous and somewhat

mysterious structure which grows by information processing. This idea was pur-

sued by the location theorists who conceived knowledge as a storage medium for

the processed information (Gueldenberg, 2001:160). But this theory is outdated

(compare Gueldenberg, 2001:161 and his notion of the ‘old’ location theory) as

this level of comprehension could already be left behind by recent advances in

neural biology (cf. Anderson, 1995 and chapter 2.1.1). For example, the cognitive

scientist Sodian (1986) notes that cognitive structures are not the storage for

knowledge, they are knowledge. This moves the discussion away from the idea,

that knowledge is stored in storage media towards its locus in the connections (be-

tween the nodes). Here, Weick and Roberts (1993) confirm, that knowledge re-

sides in patterns of connections not in indivuated local symbols. Simultaneously,

this shows the difference between the terms knowledge and content as the concept

knowledge (as a system with connections) is not synonymous to the term content

(with its emphasis on resources/objects and less on structures). This rules out defi-

nitions which simply reduce knowledge as different types of contents stored in a

memory system (like Baddeley, 1997 or Squire, 1992).10 Here, a boundary be-

tween the terms knowledge and memory can be drawn. Memory is the structure

on which knowledge can be employed, with knowledge being no object but a dy-

namic structure to generate action from information (also cf. Lehner, 2000:94).

the previous paragraphs show that this is only one consistent and logically connected

perspective and can not be separated into three different types.

8 Wiegand adds a further hypothesis to that discussion. It represents the idea which is also

proposed by the theories on Organisational Learning: Collective knowledge is the

product of many individual learning processes transcending individuals (Wiegand,

1996). This work will later show, how this is supported by theories on transactive

memory systems etc. However, it is less an insight than a heuristic to manage a com-

pany in a dynamic environment

9 This notion relates to Argyris and Schoen (1996) and their two forms of learning and fur-

ther to Kim’s OADI cycle stages Assess and Design (Kim, 1993).

10 However, it has to be added that patterns can become symbols again, just like a square is

a pattern of four lines which are a pattern of dots.

Visualization of virtual Knowledge Communities 17

Luhmann says quite similarly: Memory is the apparatus which allows storing suc-

cessful experiences for a reuse (Luhmann, 1984:75).11

Further terms related to the core concept knowledge can be found in Lehner

(2000:141). They include intuition, competence, explanation, message, emotion,

proof, intelligence, or wisdom and can be added to get a clearer picture of the term

knowledge.

Figure 4: The most important Borders of the Term Knowledge.

In summary, the term knowledge can be approached by defining its borders (cf.

Figure 4; the concepts are positioned in a way that reflects their connections). Fol-

lowing this strategy, knowledge can be understood as a mechanism which isolates

information from its environment using attention as a filter. It accommodates the

information, because it is relevant for the system as it improves the knowledge to

enable a better achievement of the objective of successful actions and decisions in

order to manipulate or adapt to the environment (identity, stability, growth). By

this process the implemented information is being equipped with a meaning which

is interdependent with causality which in turn steers the attention again. Simulta-

neously, the confidence about the individual estimate about the projected success

of an action can reach from subjective believe (which could not be proved yet) to

objectively reaffirmed ‘truths’ (confirmed by the environment). The meaning is

also derived from the observations of the system’s actions. The knowledge is us-

ing the memory to work on and to store the information. The memory again is a

network structure which is able to store new incoming information (and hence is

also called a knowledge carrier) in a way that preserves consistency. This purpose-

11 Lehner brings this division into the KM discussion: The organisational memory (OM)

has the core functions acquire, store, and access to knowledge (Lehner, 2000:126)

Knowledge

Memory

Causality

Networks

Actions,

Decisions

Attention

Information

Signals

Data

Learning

Meaning

18 M.Trier

ful addition of information to improve the knowledge of the system is also called

learning.

For better applicability, this definition can be simplified as follows: Knowledge is

a network of interpreted signals and data from the environment, which is continu-

ously reorganized and made consistent by the processes of learning and thinking

in order to derive and infer promising decisions and interactions for manipulating

and adapting to the environment.

The approach to define knowledge by means of defining its borders requires a dis-

cussion of its generation, usages, and meaning. It thus yields in the conceptualiza-

tion of knowledge as a residue, which remains after deducting all related terms

around it. Although it renders knowledge an underlying foundation of all the other

concepts, this leaves the actual term knowledge as a black box. However, a first

important implication for the management of knowledge can be derived: By man-

aging the processes of processing and connecting information as well as its rela-

tion to action and objectives, knowledge is indirectly positively affected.

A second strand of discussion tries to strategically approach the term knowledge

by breaking it down into classifications of types. The idea is that by this proce-

dure, the term gets more operative and beneficial interventions to positively affect

knowledge can better be derived.

The first separation into knowledge types in the more recent literature can be

traced back to Barnard (1938:235). He differentiated knowledge which is logically

explicable by language and related to behavior versus irrational behavioral and

non-verbalizable knowledge. However, this distinction later became famous under

a different name. In 1966, coming from cognitive sciences Polanyi divided tacit

versus explicit knowledge (Polanyi, 1985). This basic dichotomy later became the

most important distinction in modern Knowledge Management approaches (com-

pare Nonaka and Takeuchi, 1995 or Hansen et al., 1999).

A different type of classification includes categories which relate to the require-

ments of an organization and its employment of knowledge. Sackmann (1992) dif-

ferentiates between dictionary knowledge of facts, directory knowledge of proc-

esses and actions, further recipe knowledge (shoulds) including rules and problem

solving strategies, and axiomatic knowledge (why) of basic assumptions or cause

and effect relations (cf. Gueldenberg, 2001:187). The concept of factual12 diction-

ary knowledge is very close to Polanyi’s explicit knowledge and Sackmann’s pro-

12 Interestingly Willke says, that declarative knowledge of facts is no useful cate-

gory as is it the same as data (Willke, 1998:12). However, this does not recognize

that the knowledge of the fact being true is the actual value of fact knowledge.

Moreover, fact knowledge influences the framework of relevancies

(sense/meaning, cf. Luhmann, 1984) to select information from the environment.

Visualization of virtual Knowledge Communities 19

cedural directory knowledge is hard to transfer and hence resembles tacit knowl-

edge. The last two types are more or less abstract types derived from the first two

by logical reasoning. For example, if one knows the procedures for swimming,

one can look at how it could be done more efficient and why this is so.

Another approach was proposed by Squire (1992). It is a bit more efficient and

more detailed than Sackmann’s. Squire classifies into declarative knowledge,

which in turn consists of episodic knowledge about past events and of semantic

knowledge of facts and concepts. Further, there is procedural knowledge about

abilities, prepositions, and tacit elements. Finally he assumes meta-knowledge for

planning and controlling actions (also cf. Lehner, 2000:84). So obviously the sepa-

ration of facts versus procedures has been maintained and the more abstract

types have been integrated into one type, thus making this structure easier. This

trinity has later been reduced even more to the important distinction into factual

and methodical knowledge or simply know-what and know-how (cf. Wiegand,

1996). However, Kim (1993) later substantiates the importance of the level of ab-

straction. He separates concrete routine knowledge and abstract framework

knowledge and analyzed their relation to learning processes (Lehner, 2000:134).

Another interesting source of classifications is the Association of German Engi-

neers (VDI). Already in the 1992, it preempted various dichotomies in its guide-

line 5007. It separated explicit versus tacit knowledge (taken from Polanyi), plan-

ning versus experiential knowledge (which is similar to theoretical and assumptive

knowledge vs. practical experience), subjective versus objective knowledge

(which is subjectively true versus inter-subjectively being confirmed as true), and

logical rational insights versus fuzzy knowledge (which includes different levels

of abstraction; cf. Lehner 2000:146). This notion of fuzzy or soft knowledge has

later been extended by Rao and Goldman-Segall (1995) which utilize a spectrum

from hard to soft to illustrate the level of abstraction and its role as a primary dif-

ferentiator for different knowledge types. They differentiate hard and concrete

data – semi-concrete know-how about technology, business rules and procedures,

semi-abstract stories and myths, and finally abstract and soft knowledge about

roles, structures and culture. In their approach, Rao and Goldman-Segall relate

factual and process knowledge to the levels of abstraction proposed by Kim

(1993) three years earlier.

The knowledge types introduced so far have been extended by the idea of knowl-

edge in a group structure (remember Krogh/Roos (1996) and their definition of

knowledge in a network of people). Following this idea, Schneider (1996) defines

a mixture of explicit and tacit knowledge in a relationship network as collective

knowledge and proposes it as an important source for competitive advantage as it

is difficult to imitate and copy. To create collective knowledge continuous conver-

sions between knowledge types (cf. Nonaka’s model of organizational knowledge

creation, 1995) are necessary. Combining the basic separation into tacit and ex-

plicit knowledge with a group context yielded the definitions of further types of

knowledge as proposed by Wiegand (1996). Group knowledge can be explicit in-

20 M.Trier

dividual knowledge, tacit individual knowledge, explicit or explicable group

knowledge, tacit group knowledge, and explicit vs. tacit individual knowledge

which is dependent on group knowledge (cf. Lehner, 2000:119).

A very topic oriented categorization comes from Willke (Willke, 1995:330). He

mixes factual and methodical knowledge types with knowledge about various top-

ics and separates factual knowledge about structures, social knowledge about per-

sons, time-related knowledge about processes (timing and synchronization), op-

erative knowledge about projects (procedures) and cognitive steering and planning

knowledge. With these types, he identifies different tasks (or dimensions as he

calls it) for the management of knowledge.

Figure 5: Approaching the Term Knowledge by segregating Sub-types.

North (1999:61) separates further categories by looking at the different values of

knowledge. First he describes a connection between the types planning versus ex-

periential knowledge and subjective versus objective knowledge. They all relate to

the scope of validity of the knowledge: subjective, objective, or objectively valid

under defined conditions, which is similar to scientific approved knowledge. This

is simultaneously related to the degrees of confidence in a concept to be true. This

results in the separation between meaning, believing, knowing and inter-subjective

affirmation.

The level of abstraction is used again for the differentiation between generic ver-

sus specific knowledge. Other interesting properties which not directly necessitate

a dichotomy are the durability of knowledge (as the duration of its validity), and

its uniqueness (or exclusiveness) versus commonness (North, 1999:61). The ex-

Knowledge

Memory

Causality

Networks

Actions,

Decisions

Attention

Information

Signals

Data

Learning

Meaning

explicit vs. tacit

facts vs.

methods

concrete vs. abstract

theoretical vs. Approved experiental

rational vs. fuzzy

hard vs. soft

knowledge as a residue

types of knowledge

Visualization of virtual Knowledge Communities 21

clusiveness is not a direct property of knowledge but can be attributed if it is

known who has knowledge about this issue.

All these approaches should not be seen as concurrent ideas; rather they conjointly

define a set of properties which closer define the term knowledge (also compare

Lehner, 2000:140). Following these approaches, knowledge can have qualitative

properties like a high or low level of abstraction or validity. Further, it has con-

tent-related properties like a relation to facts, structures or methods, and finally

there are hypotheses about looking at its occurrence in a group or in an individual

(also compare Figure 5). Figure 6 gives a final impression about the variety of

categories which can be employed to differentiate knowledge types.

Figure 6: List of Knowledge Types. Source: Strube et al. (1996)

Summarizing, knowledge can be defined as the ability to utilize a learned network

structure of information about facts and procedures in order to draw inferences on

A very comprehensive collection of knowledge types has been proposed by

(Strube et al., 1996:799). The authors differentiate:

• General knowledge

• Common sense knowledge (culture-specific acquired during socialization)

• Analogical knowledge (for example imagery)

• Assertional knowledge (which is simply fact or the case)

• Conceptual knowledge (knowledge about the meaning of concepts and terms)

• Domain-specific (expert) knowledge (expertise)

• Episodic knowledge (related to specific cases)

• Explicit knowledge (conscious knowledge which can be documented or told)

• Erroneous or inappropriate knowledge (wrong assumptions, not yet verified)

• General knowledge (knowledge with unlimited validity in a certain domain)

• Heuristic knowledge (gives orientation for selecting alternatives without proof)

• Background knowledge

• Implicit knowledge (unconscious knowledge which obviously exists)

• Causal knowledge (knowledge about the relation between cause and effect)

• Control (or strategic) knowledge (knowledge which helps to provide orientation in a

task sequence)

• Conceptual knowledge

• Meta-knowledge (knowledge about the application of other knowledge)

• Normative knowledge (norms)

• Problem-solving knowledge

• Propositional knowledge (knowledge represented by propositions)

• Procedural knowledge (knowledge about actions, sequences and movements)

• Qualitative (and quantitative) knowledge (knowledge about qualitative relationships

• Spatial knowledge (knowledge about the spatial arrangement of objects

• Rule-based knowledge (general procedural knowledge)

• Schematic knowledge (knowledge related to structures)

• Semantic (or terminological) knowledge (knowledge about concept hierarchies)

• Default knowledge (general standards, which sometimes have exceptions)

• Temporal knowledge (knowledge about the temporal sequence and order of events and

states)

• Uncertain knowledge (includes the propability of being inaccurate)

• Incomplete knowledge (missing pieces)

• World knowledge (non-explicable meta-knowledge not related to language)

22 M.Trier

various levels of abstractions which help to plan and initiate successful actions

with a certain level of confidence. Overly short, this could be reduced to knowl-

edge as the ability to combine and apply a multitude of information to generate

solutions.

2.2 Reviewing existing KM Approaches

After having introduced a working definition of the term knowledge, the aspect of

an active and supporting system of interventions to foster and develop appropriate

knowledge in an enterprise can be discussed. Here, the literature on Knowledge

Management (KM) suggests a plethora of approaches. Subsequently, a reader who

is interested in learning about KM often finds himself completely helpless, when

faced with hundreds of mostly unrelated or even unsubstantiated ideas and con-

cepts. The emergence of the research field was accompanied by a high eagerness

to explain the complete perspective in very abstract approaches (usually showing

boxes representing domains and relations between them). This completely under-

estimated the complexity of the endeavor. In the beginning, these approaches were

very inconsistent and incommensurable. However, overtime, slowly a selection

process started and some useful ideas survived. This section will now briefly re-

view the most important approaches in the field with a special focus on European

concepts. As outlined above, the event of the discipline can be separated into two

major waves. The first one could be described as ‘finding identity’ and the second

as ‘finalizing structures and useful analysis approaches’. However, even before

these two waves, preliminary concepts slowly fueled the idea of a knowledge ori-

ented perspective on enterprises.

2.2.1 Preliminary Concepts related to modern Approaches of KM

The idea of managing knowledge in a company is not new. The first implicit ap-

proach to deal with this issue in a corporate setting has indeed been pursued by

Frederick W. Taylor as early as in 1911 (Taylor, 1983:38). In his scientific man-

agement approach, he managed his workers knowledge in a special way. His man-

agers observed the worker’s actions and tacit experiences and explicated job de-

scriptions from it to create standardized methods. However, it has not been treated

as a source for knowledge generation and management. Further, it has not been

included in conceptual models or in the creation of dynamic learning processes

(cf. Pawlowski, 1994:16). Thus, Taylor did not recognize his workforce as a con-

tinuous source of new knowledge (cf. Pawlowski, 1994:16). Later in 1959, Edith

P. Penrose recognized that it are not the resources provided for a production proc-

ess but the services applied to the resources that matter. Such services are a func-

tion of accumulated experience and knowledge in an enterprise (Penrose, 1959).

She was thus the first who saw the relation between experience, knowledge, ser-

vices (as actions), and products. Still, she observed the issue but did not propose a

Visualization of virtual Knowledge Communities 23

theory to support a process, which supports the generation and reuse of experience

or knowledge (Nonaka and Takeuchi, 1997:48). However, she implicitly antici-

pated the perspective of very modern approaches to Knowledge Management

(KM) which aim to provide knowledge to support business processes (see section

2.2.3 on second wave approaches to KM).

Another pioneering line of thought was provided by Peter Drucker (1993). In the

1970s, he first talked about the related vocabulary including Knowledge Work,

knowledge worker and Knowledge Society. Together with the concepts of a

knowledge company proposed by Nonaka (1991), the topic of Knowledge Man-

agement emerged. Although, these two works were mainly responsible for the

emergence of KM as a discipline, approaches from a multitude of disciplines re-

lated to the philosophical (and later economical) core concept of knowledge and

knowing have influenced the emergence of KM. All these fields are necessary to

consider in order to fully capture the rich body of theories subsumed under the in-

tegrative and inter-disciplinary umbrella of the new approach Knowledge Man-

agement. This multitude of perspectives and prior theories eventually renders the

management of knowledge a truly integrated and multi-disciplinary approach.

However, it has to be noted, that the claim to provide a holistic and complete KM

approach which recognizes all underlying theories is conflicting with the prag-

matic objective of corporate applicability. In order to understand the breadth of

theoretical roots and the background of KM, the related fields and their connec-

tions will be very briefly introduced now. However, as each of the disciplines

mentioned would require comprehensive introductions, this section will only give

a first overview in the form of a list of concepts and their main relations. It is up to

the reader to take this list as a starting point for further inquiry into the different

preceding subjects.

The fundamental sciences related to KM include epistemology, which is the phi-

losophy of consciousness and insight, and cognitive psychology, which analyzes

the psychological learning processes and autopoiesis. Both of these fields relate to

the previous sections. Epistemology tries to describe the concept of knowledge

and insight to answer the question, how to achieve expertise from learning using

philosophical theories and concepts whereas cognitive psychology approaches

these two issues by examining the human apparatus of cognition using psycho-

logical experiments. From that basis constructivism emerged as a special instance

of epistemology, which claims, that all insight is constructed by an individual.

There is no objective ontology. That also influenced the branch of cognitive psy-

chology which subsequently developed towards a research field called connection-

ism. It explains how the brain structure can perceive and think by using its con-

nections of neurons. At a different level of observation, sociology analyzes social

systems (which includes organizations and hence corporations). More recently so-

ciological research has been influenced by systems theory. The latter serves as the

general underlying meta theory and also allows connecting sociology with episte-

mology.

24 M.Trier

Another very important preceding research area of KM was the theory of Organ-

izational Learning. It combines the introduced individualistic (epistemological)

and the organizational (sociological) perspective and specializes on analyzing cor-

porations. Primarily, Organizational Learning transfers and combines existing

ideas and insights from other disciplines like autopoiesis from systems theory,

learning from epistemology and cognitive psychology, or group processes from

sociology and systems theory. They are employed for the analysis of learning

processes in organizations (without actually establishing a novel theory). This ap-

plication usually resulted in programmatic and normative suggestions (cf. Senge,

1990) or checklists for ideal processes and structures which connect learning to

corporate success (cf. Krallmann et al., 1999:304). However, this branch of re-

search reinvigorated the organic and sociological perspective as a plausible per-

spective for analyzing enterprises. In connection with engineering sciences and

especially with computer science, the theories and approaches of Knowledge

Management (which are discussed in more detail in the next chapter) extend these

concepts of Organizational Learning on a more operational and process-based

level with their focus on proposing improved technical support and implementa-

tion. This is also resulting in a strong link to another root of KM, namely Informa-

tion Management and its technologies. A main contribution of the new KM ap-

proaches was it to develop a content-related perspective on the individual and or-

ganizational learning processes in companies (the artifacts that document and

hence ‘include’ knowledge). The management of a multitude of knowledge-

related content artifacts employs methods of Artificial Intelligence which deal

with structures of knowledge-related entities. Here, the ongoing challenge of

Knowledge Management is to combine the classical and complex contrast be-

tween engineering and human sciences. However, most KM approaches have no

feasible approach for combining these two perspectives and therefore either deal

with technological problems (like product oriented KM, Organizational Memory,

or Information Retrieval), or with people-related issues (like process or human-

capital related KM which is often very similar to Organizational Learning). How-

ever, this second branch is more capable of clearly differentiating itself from In-

formation Management, which nevertheless serves as a basis science. KM con-

cepts aiming at a true integration of both perspectives are usually only offering

and discussing methods for a strategical level and are unable to be transferred to

everyday operations.

Figure 7 summarizes the numerous disciplines and their joint contribution to the

emergence of KM as a field of research. Please note, that the description of the

fields has been abridged and simplified to enable for a maximum of transparency.

In general, the six boxes on the right hand side comprise the technological influ-

ences on KM whereas the remaining boxes on the left hand side represent the peo-

ple-oriented sociological impact.

Visualization of virtual Knowledge Communities 25

After having introduced the preliminary influences and ideas which resulted in the

distinctive discipline of KM, now the actual development of this field of research

will be examined in more detail.

Figure 7: Overview about the Roots of Knowledge Management and their Connections.

Before the next chapter starts to introduce the approaches of the first wave of KM,

it has to be noted, that generally, it is very hard to compare and summarize the ap-

proaches. Therefore, this chapter tries to identify the connections between the

most influential concepts by chronologically arranging them on a timeline and by

analyzing their scientific contribution to rendering a consistent theory and to de-

velop applicable instruments. Here, it has to be warned, that the act of reconstruct-

ing a timeline is strongly subjective as it is subject to the author’s focus and the

selection of the ‘main’ idea or contribution. The primary problem is, that theories

do not take into account each other completely, which is sometimes also intended

to challenge existing theories. This results in heterogeneity and even incom-

mensurability of new ideas which are not discussing all previous thoughts. More-

over, it is usually difficult to understand the tacit parts of many theories (mainly

because the tacit understanding of the author’s research object ‘knowledge’ can

differ).

2.2.2 The first Wave of KM Concepts – Push Orientation

As already anticipated in the introduction to this section, remarkably in line with

the economic cycle, Knowledge Management came about in two waves. The first

wave went until about 2001. Following the general zeitgeist of the new economy,

KM researchers and practitioners wildly proposed a multitude of approaches and

opinions about how KM should look like. This left the student of the literature

very confused and without a clear picture of how to employ KM. In this situation,

Epistemology

(Philosophy of human

understanding and insight)

Systems Theory

(Structure and behavior of

General Systems-

Meta-Theory)

Behaviorism

(Theory of human

understanding

and insight )

Constructivism

(Theory of human

understanding

and insight )

Sociology

(Theory of social

Systems )

Informatics

(Special Science including

Math, Logic, Physics)

Psychology

(as empirical study of

behavior)

Cognitive Psych.

(as empirical study of

Understanding and insight)

Theories of Action

(analyzing behavior

in spec. situations)

Artificial Intelligence

and Neuro-Infor-

matics

Connectionism

(as neural model of cognition)

Self-Organization

(Chaos, Autopoiesis)

Organizational

Learning

(combining indiv. and org.)

Organizational

Memory

(IT systems as metaphor

for memory in a corporation)

Economics

(Effectivity, Corporation as

Organization, Management,

Objectives)

Knowledge Mgmt.

(integrating OL processes,

OM structures, content arte-

facts and knowledge carriers)

Personnel

Development

(Qualification,

Skill Management)

Organizat.Change

(Change Management,

Innovation Management)

26 M.Trier

risk friendly companies competed for being the first to have implemented the lat-

est idea without actually checking feasibility and barriers beforehand. Some

smaller companies even superficially implemented KM systems only for their

communication strategy towards their shareholders without really employing the

systems. However, this quick practical adoption resulted in a steep learning curve

and a fast progress of theory development.

This book’s approach to classify and evaluate the approaches is to position them

in a way that highlights their contribution to the overall objective of the theory of

KM: to generate an operative set of measures and instruments for corporate appli-

cation, which are derived from a general and appropriate KM vision.

The following Figure 8 shows the employed roadmap concept for analyzing KM

theory development in more detail. The applied scheme includes seven stages. It

assumes that KM approaches advance by subsequently aiming at achieving two

subsequent objectives. They first seek to precisely describe the problem, which

KM is about to solve and hence relate to the question ‘WHAT has to be done by

KM?’ – thereby pointing to the problem. In a second step, the discipline pro-

gresses towards proposing answers to the question of ‘HOW can it be done?’

which points to the (practical) solution. On every step, hypotheses on different

levels of abstraction can be examined to advance the comprehension of an other-

wise very unstructured evolvement of KM theory.

Figure 8: A generic Roadmap for the Development of KM as a Theory.

The broadest level of KM-related theoretical contributions includes approaches

that discuss the vision for Knowledge Management (stage 1a) and the boundaries

(1b) of the discipline. Here, the introduced preceding roots are discussed and ex-

tended. This gives a first high-level orientation for the overall research direction.

Afterwards, on a more advanced level, authors worked (and continuously work)

on a shared definition (1c) of what KM actually is. Using this definition, further

Tasks

Abstract

and

Theoretical

Manage-

ment

Interventions

Strategic

Manage-

ment

Interven-

tions

Operative

Measures

and

Instru-

ments

Require-

ments

Objec-

tives

Vision

Boundaries

Definition

of K

and KM

HOW - Pointing to the SOLUTIONWHAT --- Pointing to the PROBLEM

Looking at knowledge types

Looking at the generic term knowledge

Looking at the existing organization (Process, Skill, Communication, Document view)

Looking at knowledge flows between types and people

Looking at the existing organization (8 Blocks strategic view)

Visualization of virtual Knowledge Communities 27

approaches can discuss and propose the objectives (2) of managing knowledge

and the respective requirements (3), which have to be fulfilled when Knowledge

Management is to be introduced in a company. The most detailed level of theory

evolvement is an elaborated and theoretically sound discussion of the actual tasks

of a (holistic) management system for knowledge (4). However, up to this point,

the according contributions have only clarified the problem but have made no

suggestions for how the actual management of knowledge should look like in

practice. This subsequent solution oriented domain of KM theory starts with first

approaches towards abstract and theoretical management interventions (5), fol-

lowed by a more detailed description of strategic management activities (6). The

most advanced concepts target the derivation of practically applicable operative

measures and instruments (7), which have of course to be based on all other stages

of KM theory development. This whole sequence of evolutionary steps (1 to 7)

can be perceived as a top-down approach to developing KM theory. These steps

will now be utilized to classify and allocate the various introduced KM theories of

the first wave in order to create transparency about the underlying progress in the-

ory development.

As introduced in the previous section, a first major inspiration for KM theory was

provided by the preceding Organizational Learning theory13. It analyzed corporate

learning processes and as a by-product discovered the idea of managing the results

of learning processes: experience, memory (storing experiences) and finally

knowledge which can be employed using this memory14. For example, Hedberg

analyzed in 1981 (Hedberg, 1991) the concept of Organizational Memory. Ac-

cording to him, it determines the cognitive processes of information processing for

the whole organization, which constitutes the structural basis for Organizational

Learning processes. He creates the link between the processes of learning, the or-

ganizational memory structure which is utilizing learning processes as input, and

knowledge which is needed to subsequently utilize the Organizational Memory.

This concept of Organizational Memory has been analyzed more exactly by

Wegner in 1987 who suggested the concept of a Transactive Memory System,

which is a mutual dependent system of knowledge storage. People scan individual

memory using communication to socially combine found elements (Wegner,

1987:191). According to the classification scheme proposed in Figure 8, these first

contributions tried to render a rough description of the concepts of knowledge and

of Knowledge Management but only as a by-product. They approached this

topic’s borders from the outside and framed a first rough vision of a discipline,

which later should evolve to become KM.

In 1989, Pautzke (1989) focused directly on the element knowledge and isolated it

from the learning processes and memory structures in order to analyze its proper-

13 Compare for previous section for a very brief overview about the sequential relations of

preceding related theories.

14 Also compare the concepts related to the term knowledge in chapter 2.1.3.

28 M.Trier

ties in more detail. This description led to the segregation of different layers of

knowledge, reaching from accessible individual knowledge to inaccessible indi-

vidual and collective knowledge. Although he creates a first typology of the or-

ganization’s knowledge base as the knowledge, which can be utilized by the or-

ganization, the model is very abstract and of low utility for corporate practice as it

does not allow for deriving systematic ways of how to employ this asset. Still, he

contributed to the later discussion of knowledge types (also cf. section 2.1.3) and

their role for different KM activities and he directed the focus to the abstract po-

tential (which relates to the vision) of trying to manage the knowledge in a com-

pany.

Also in 1989, Kleinhans conceives knowledge to be managed as an object which

can either exist directly or can be accessed and isolated via the knowledge carriers

(as a ‘container’ subject). This perspective on knowledge does not take into ac-

count the complexity of the concept of knowledge as defined by epistemology.

Still, this reductionism enabled Kleinhans to determine a technical, personnel, and

an institutional problem domain for the field of KM. This is emphasizing the mul-

tiple relations to Human Resource Management and Organizational Science. He

also includes Computer Science, thus identifying the potential of extending the

related concept of Information Management. In this way, Kleinhans outlines a

boundary of the discipline KM and gives a first and very rudimentary outlook on

its objective. Subsequently, researchers in these disciplines recognized the dis-

cussed potentials, implicitly adopted this pragmatic perspective, and started re-

searching knowledge and its management in their disciplines.

One early example of how this momentum influenced other disciplines is the con-

ceptualization of different application layers of Information Technology by Han-

ker in 1990. He supposes the four layers: support of operative processes, manage-

ment support, competitive strategy, and finally organizational strategy including

Knowledge Management as the highest level (Hanker, 1990). This also shows that

KM initially has not been positioned at the operational end.

By analyzing the structure of Organizational Memory in a company, Walsh and

Ungson (1991:64) are also indirectly contributing to the field of KM by extending

and operationalizing the concepts of the above authors. They identify six different

knowledge carriers (i.e. repositories): people (which store experience using tech-

nical support and knowledge carriers), culture (storing values and soft issues),

transformations including processes and individual activities, structures (and the

according roles), ecology (workplace), and external archives. Although the authors

explicitly focus on relevant information and its locus, they also analyze how the

information is stored and retrieved from a ‘retention facility’. They outline a pre-

liminary picture showing which knowledge domains have to be managed in a

company, implying the big scope of this endeavor. Next to helping to define broad

domains for a KM initiative, i.e. working on culture, people, and processes, they

see the actual artifacts as only indirectly accessible results of people’s interactions,

Visualization of virtual Knowledge Communities 29

or cultural and business processes. This results in a clearer picture of the theoreti-

cal boundary of KM and its major objectives, as the authors suggest, where KM

has to intervene.

An early and very influential concept of how to manage knowledge also rests on a

typology of knowledge. It was produced in the same year by Nonaka (1991)15.

Nonaka did not work with locational properties of different knowledge domains,

but with the properties of different knowledge types. By this, he directly relates to

previous epistemological research by applying Polanyi’s (1985) separation into

tacit and explicit knowledge. He is relating this concept to the individual level and

much more important, he links it with the Organizational Learning perspective

when he analyzes knowledge creation at the individual and organizational level.

Here, he identified four important types of knowledge transfer, namely Socializa-

tion, Externalization, Combination, and Internalization16. Nonaka also manages to

provide a first empirical validation of a KM concept (which is up to now not pro-

vided by many ‘approaches’ in the field) using an analysis of product and process

development in Japanese enterprises.

With this approach, Nonaka’s primary theoretical achievement was to shift the re-

search focus away from Walsh and Ungson’s restrictive conception of knowledge

as an object in a retention facility towards the flow, conversion, and transfer be-

tween different forms of knowledge on different organizational levels. This

change in perspective has the advantage to not limit Knowledge Management to

storing artifacts but to look for the actual transfer, communication, and develop-

ment of knowledge in social processes between people. This also strengthens the

position of conceiving knowledge as bound to people. Next to bringing KM theory

in compliance with the connectionist discovery that knowledge is a network struc-

ture of interrelated information (neural networks), the actual challenge of KM is

hence less the systematical storage which can be left to Information Management

but rather the utilization of the domain of tacit knowledge. The lack of a practical

implementation shows that Nonaka’s approach does not provide a real operative

and prescriptive procedure for improving Knowledge Management in a company

but it aids as a preliminary think tool. However, from the perspective of KM the-

ory advancement, the division into knowledge types and the focus on the flows

between them influenced the definition of the scope and the objectives of the dis-

cipline very much. Nonaka furthermore contributed to the analysis of objectives

and requirements for knowledge transfer, when he suggests five prerequisites for

knowledge generation in organizations: intention (vision), autonomy of employ-

ees, fluctuation (dynamics), redundancy, and internal heterogeneity. The move-

15 However, the most complete work was the co-production of Nonaka and Takeuchi pub-

lished in 1995.

16 This is why this model is also referred to as SECI-Model. This work will not describe

such theories in more detail, as a plethora of descriptions can be found in other docu-

ments, including the internet.

30 M.Trier

ment towards such a corporate environment is a first strategic objective of

knowledge oriented management. In the years 1992 to 1995, Nonaka went to ex-

tend his first paper to a book and many concurrent approaches tried to build on his

ideas.

Somewhat not incorporating this development, in 1993, Albrecht (1993) framed a

second trinity of domains next to the one of Kleinhans in 1989. It consists of

Knowledge Resource Management, Human Resource Management (HRM), and

Knowledge Technology Management. He managed to propose a first list of ge-

neric tasks, which can be attributed to the discipline KM (Albrecht, 1993:102).

They included the combination of knowledge orientation and strategic vision, the

creation of a knowledge-oriented corporate culture, the development of a knowl-

edge strategy, the strategic management of human resources and of knowledge

oriented technology as well as the practical implementation of the knowledge re-

lated strategy. However, there is no concrete suggestion of what these steps con-

tain and how they should be executed. Still, this list of tasks also shows that a

knowledge manager has no novel and distinct field of expert work, but rather inte-

grates existing perspectives like HRM, IT, Strategy, and Leadership and empha-

sizes their important interconnections.

In the year 1996, the momentum of the discipline increased again. The existing

rudimentary theoretical foundation for the objectives of KM was improved by the

work of Grant (1996), who explored the resource based view of KM and found an

economical foundation for the role of KM, which also substantiates the impor-

tance of this discipline for management. According to Grant, enterprises exist be-

cause of the limited capacity and capability of the human brain to acquire, store

and process knowledge. This results in specialization, which requires coordination

to solve complex solutions. Markets are unable to manage this coordination, be-

cause they can not mobilize tacit knowledge. That is why enterprises allow indi-

viduals to integrate their special knowledge. KM is therefore the creation of condi-

tions which allow the employees to build collective knowledge.

Unsatisfied with the many normative but unproved hypotheses, assertions and in-

applicable KM objectives which have been proposed in the first half of the 1990s,

Schneider (1996:34) worked towards more operative requirements of KM. He

identifies three necessary shifts in theory:

• from concentrating on ideas and proposals towards supporting people and

their ability to generate ideas and proposals,

• from employing documentation and data storage towards reuse and de-

velop data in interpersonal networks, and

• from informed experts towards networked and shared learning.

More generally, he proposes a concept with the elements combining existing

knowledge, generating new knowledge, improved documentation, and systematic

Visualization of virtual Knowledge Communities 31

observation of the environment to increase knowledge. He furthermore stresses

the importance of capturing the economical benefits and costs to create a control-

ling and management around these tasks. Schneider differentiated the three ge-

neric domains technology, organization and employee and derived a list of more

operational tasks for each domain, including connecting and integrating IT, de-

creasing organizational hierarchies, hiring diverse people, awarding creative ex-

periments, assigning free budgets, emphasizing teams and empowerment, creating

a learning culture, implementing mentor programs or yellow pages, and develop-

ing trust among the employees. On a very abstract level it can be observed that all

his tasks relate somehow to networking. This first implicit anticipation of a very

fundamental underlying mechanism of supporting Knowledge Work will be later

discussed and elevated to become the main perspective of this book and its devel-

oped contribution to Knowledge Management.

Schneider’s concept is a good example of how theories began to transcend the

academic and unprogressive discussion of the problem domain and moved to-

wards a solution perspective with first generic suggestions of ‘how’ to do KM.

A clear indication of this development has also been given by Reinhardt and

Beyer (1997), who propose a set of questions for a structured KM Analysis: How

well is information identified and acquired? Is tacit knowledge sufficiently being

provided to others? Is the system of communication channels appropriate or are

there barriers for communication? Which assumptions determine activities? Is

there openness to change and diversity? Together, these questions provide a first

guidance for practical implementation of a KM analysis but miss the necessary

underlying procedure and methodology to comprise a theoretical KM approach in

its own right. However, the authors’ contribution somehow preempts the strong

analytical touch and (inverted) bottom-up-approach of KM found in the second

wave concepts, discussed in the next chapter.

Schueppel (1996:192) suggests a further but more methodical generic analysis. It

starts with the determination of ‘knowledge elements’ and analyzes individual and

collective learning processes around these elements. Then, barriers for these learn-

ing processes need to be identified in order to derive appropriate changes of the

organization. Although this concept starts by looking at knowledge elements (top-

ics), it is very generic and does not systematically relate the identified problems to

appropriate solutions. Hence, it is not providing much operative guidance for KM

projects as it is not much related to the everyday operations and work tasks. How-

ever, the perspective of analyzing learning processes again highlights the strong

relation between KM and Organizational Learning.

Another approach towards KM domains which stresses the connection of learning

and KM was suggested by Willke (1996). He claims that the creation of an intelli-

gent organization rests on concepts for competence management, qualification,

and the ability to learn. To some extend, this approach is similar to Human Re-

source Management but uses a new terminology, thus its insight is similar to that

32 M.Trier

of Schneider who published his concept in the same year. The two previous ap-

proaches of Willke and Schueppel imply the difficulty to operationalize the exist-

ing (organizational) learning theory to support operational business tasks.

A contrary example, which is very operative but hardly includes people or learn-

ing, is the rather technical Lifecycle Model of KM by Rehaeuser and Krcmar

(1996:20). It suggests abstract tasks which include the management of sources

(identify, generate, and capture), of knowledge carriers (storage) and resources

(artifacts), knowledge supply (provision of knowledge to problems, refinement of

knowledge), knowledge demand (interpret supplied knowledge and act), as well as

communication and processing infrastructures (personnel, organizational, techni-

cal). The model is descriptive, but not proposing directions for actions (North,

1999:167). Its focus on explicit knowledge allows for a very concrete approach

and shows that concepts for documentation and IT-related tasks of KM are far eas-

ier to establish than strategies targeting tacit knowledge. However, it is actually

not meeting the core challenge of KM, because if the people oriented perspective

(i.e. tacit and collective knowledge) and the learning processes are excluded, this

perspective does not exceed the conventional idea of information management.

Analyzing the two different strands of concepts implied by the previous contribu-

tions, Schueppel (1997:187) could categorize two generic families of approaches

which focus two domains: human oriented and technology oriented KM. Driven

by sociological and psychological insights, the first defines individuals as the cen-

tral knowledge carrier and aligns KM close to Personnel Management. The latter

is only focusing explicit knowledge which should actually be called information.17

He stresses the importance of combining human capabilities with supportive tech-

nology into an integrative approach. KM theories which belong to the realm of

Organizational Learning need to merge with information management instruments

towards an integrated Knowledge Management. This integrative approach is nec-

essary to move towards the next step in KM theory advancement, the proposition

of a set of practical management activities, which utilize appropriate instruments

and technologies. It eventually prepares for scientific answers to the question of

how to arrive at a KM solution.

A further important insight of the approaches after 1996 was a central knowledge

(management) process, from which KM measures could be derived. It was

found, that the concepts can only become more operational, if the activities are

more specifically oriented towards special aspects of a knowledge flow. This

strategy is based on the preliminary knowledge process found in Nonaka’s knowl-

edge transformation approach. The concept has now become extended to opera-

tionalize the set of KM measures.

17 How could the ability to connect information and to apply it to solve a problem with a

promising action be explicated? Also compare for this book’s working definition of

knowledge on page 21.

Visualization of virtual Knowledge Communities 33

Christmann-Jacoby’s and Maas’ (1997:23) conclusion, that KM is the process of

knowledge acquisition, finding, and structuring contains a first segregation into a

knowledge process, although they do more relate to Organizational Learning than

to KM. Reinmann-Rothmeier and Mandl (1997:22) also proposed general KM ac-

tivities. They include information dissemination, information selection, the crea-

tion of knowledge networks, storing knowledge, distributing knowledge, exchang-

ing knowledge, applying knowledge, and evaluating actions to create new knowl-

edge. With these KM activities they prepare for a very detailed model of a knowl-

edge process, the renowned ‘Eight Building Blocks Model’ of Probst et al. (1997).

Here, the authors utilize the knowledge related activities to break down the

knowledge flow into six basic elements in order to move towards more opera-

tional interventions which are directed at improving the individual elements. The

heuristic is that when KM supports each building block, the overall knowledge

flow improves. The six elements are knowledge identification, knowledge acquisi-

tion, knowledge creation, knowledge dissemination, knowledge utilization, and

knowledge storage. The authors embedded these elements in a feedback loop of

knowledge planning and controlling, thereby constituting the actual management

layer of KM. However, North identifies only little support for implementation and

a missing link to operative processes (North 1999:167). An interesting aspect of

this model is that it has been derived on a forum on Organizational Learning to-

gether with many practitioners. The knowledge process elements resemble learn-

ing phases. Compare for example Pawlowski’s (1994:21) identification, genera-

tion, diffusion, integration, modification, and action. Again, the transition from

Organizational Learning to KM and the addition of IT aspects can be identified.

The model of Probst et al. (1997) has been generated deductively by analyzing

practical requirements and working towards a KM vision. This is quite contrary to

most other approaches which start by outlining the vision of KM. This bottom-up

methodology can provide for the necessary connection to business operations and

is thus a very important shift18. However, the preceding theoretical contributions

where also necessary to establish a vision of what should actually be observed and

analyzed in practice. Further, the bottom-up direction of Probst et al. raises the

question, whether the different captured aspects are able to be integrated into a

consistent project or KM vision.

A further concept related to the knowledge process is called Framework of inte-

grated KM and was proposed by Reinhardt and Pawlowski (1997). The authors

employ a learning cycle model to identify several types of KM analysis. Phase one

is called Identification and includes an analysis of recognition and information ac-

quisition. The next phase Diffusion is comprised of a communication channel

analysis plus the related communication barriers. In the Integration phase, it is

asked, whether new knowledge is ignored or integrated. The last element, the Ac-

tion stage, asks if knowledge is getting applied. This analytical concept follows a

18 (Also cf. the adversely directed arrow on the bottom of Figure 8)

34 M.Trier

qualitative approach of going into companies in order to talk about these issues in

an interview. Subsequently, change management and organizational develop-

ment measures are proposed. By this, the method very much resembles Schuep-

pel’s (1996) previously published analytical instrument which analyzed learning

barriers (see above). Further, it is a rather abstract approach which is not directly

related to the everyday work tasks (North, 1999:155). Neither there is a set of

measures related to identifying weaknesses nor are operative suggestions given for

KM implementation and improvement (North, 1999:167). Finally, the question is

how these processes differ from Organizational Learning Theories as they analyze

the processes of knowledge acquisition which is simultaneously the classical do-

main of learning research.

Probst’s bottom-up approach which included practitioner’s expertise in KM also

shows that meanwhile, the increase in practical KM initiatives allowed for the

analysis of first corporate case studies to verify the approaches, identify barriers

and observe the status quo of KM theory development and its application in enter-

prises. The increasing opportunity to actually observe KM activities in enterprises

also advanced theory development.

In an empirical study, Bullinger et al. (1997:39) identified the creation of organ-

izational networks, the methodical development of knowledge in the company,

training, and the transfer of best-practices as important project elements for Ger-

man companies. Less important than expected was storing expert knowledge.

Among the most relevant barriers the authors observed a lack of time (which is an

indicator for low priority), missing attention to the issue, missing transparency,

motivation, and transfer processes. A big potential is being projected for IT which

can be employed to integrate knowledge supply and demand systematically with a

supporting platform (Bullinger, 1998). Such empirical insights can subsequently

emphasize the important contributions (and the expected barriers) of KM instru-

ments, which needs to be taken into consideration by researchers.

One year later, Davenport (1998) reported about the most important KM project

domains for American firms and also shows practical strategies and barriers. Ac-

cording to him, in practice the term KM boils down to creating knowledge bases,

increase access to current knowledge, and to create infrastructures. This is quite

contradicting with the above findings of Bullinger et al. (1997). Davenport identi-

fied project elements which are important for KM success, namely working on or-

ganizational structure, knowledge structure, culture adaptation, integrated termi-

nology, motivation schemes, and knowledge transfer on various communication

channels. He sees the danger, that if practitioners can not systematically identify

what knowledge to work on, they tend to risk information overload by the intro-

duction of their KM technologies. Among the barriers Davenport identified miss-

ing mutual obligation, divergent contexts and cultures, low levels of trust, low ac-

ceptance of knowledge due to time restrictions or missing opportunities to meet,

and finally role constrained and superstitious learning.

Visualization of virtual Knowledge Communities 35

Next to this operative verification of KM approaches by surveying corporate ini-

tiatives, a further interesting line of thought emerged in 1998. The first event of a

process oriented view on KM can be noticed. It augments the perspective of seg-

regating the knowledge (management) process by observing the processes exe-

cuted by the experts.19

In this context, Kuehn and Abecker (1998:185) segregate a product perspective

(which looks on knowledge documents in Document Management Systems and

hence treats knowledge as an object) and a process view, which looks on social

communication processes and their technical support by groupware and workflow

management systems. Although the initial dichotomy is similar to Schueppel

(1997), it is a first approach towards a process view. However, the actual relation

to business processes was created by Wargitsch (1998:17). He demands, that KM

should look at the object knowledge and the business processes which generate it.

By this the scientific management of Taylor (1911) and the first preliminary con-

clusions of Penrose (1958; see above) finally re-enter the KM discipline.

So far, it can be generalized, that the dichotomical categories developed to better

explain the strands of KM theory obviously can be allocated to the broad concepts

‘product’ or ‘object’ and ‘processes’ or ‘communication’. On an abstract level,

this is the contrast between watching an element as knowledge or the relation be-

tween elements (which is again highlighting the initial shift in focus of Nonaka,

see above). This duality remarkably resembles early discussions in neural biology

where the theory of knowledge as a stored imagery in the brain was slowly re-

placed by the view that knowledge resides in the networks of connections between

neurons (cf. Anderson, 1995).

In 1999, the main dichotomy became empirically substantiated by Hansen et al.

The authors analyzed many corporate instantiations of KM and empirically veri-

fied the two main KM strategies, which the authors named Codification and Per-

sonalization. The underlying insight is not new: the management of knowledge

has to focus tacit knowledge using personalization and learning strategies, but for

rather standardized fields, practitioners indeed employ a codification strategy with

a strong focus on IT support. Moreover, it was found, that companies tend to focus

one of the two strategies - showing, that real integrative KM as theoretically pro-

posed for years has not reached corporate practice at that time.

The empirical findings also imply a very important further differentiation, which

can help to understand the two sides of KM as a part of an integrative approach.

Knowledge transfer can be separated into direct knowledge transfer (directly be-

tween people) and indirect knowledge transfer (indirectly via knowledge objects

or artifacts, like documents). This distinction also shows that there are actually

two process oriented perspectives possible (which is somewhat dissolving the di-

19 This later specialized towards a business process and a network oriented perspective. Al-

so compare chapter 2.2.3.

36 M.Trier

chotomy of Kuehn and Abecker, 1998:185). There is a process for direct commu-

nication between people but also one for indirectly documenting the insights of

people to internalize them afterwards. If Wargitsch’s idea is also to be included,

the question is now, how the business processes interrelate with these processes of

knowledge exchange. 20

With all the empirical insights, the discussion about required KM project ele-

ments and procedures was furthered by Seidel and Lehner (1999). They proposed

a set of project packages targeted at corporate culture, incentive systems, infor-

mation systems, and finally at leadership/management systems. This also shows

the role of IT; it only covers one quarter of a KM project.

Without really relating his work to his predecessors but by analyzing their short-

comings, North (1999) proposed a further KM approach called ‘Knowledge Mar-

ket’. He claims that it is empirically sound, because it has been developed in a

corporate project. It consists of three elements or organizational layers. First, the

framework conditions (like a vision) and the incentive system are getting defined.

Then supportive knowledge market rules are set, including the definition of clus-

ters of interests (North, 1999:233), best practices, individual experts, centers of

excellence, or networks as a visible hubs, and push vs. pull elements of the KM

initiative. Finally, from this concept, processes and structures for operative KM

are derived for the knowledge carriers person, network, process, and organiza-

tional unit, using different media. Among these operative measures are structured

processes for knowledge integration like best practice transfer or benchmarking,

coaching of knowledge development via a moderator, matching interests, and

structuring projects. Despite introducing three layers from strategic to opera-

tions, the derivation of operative instruments and their final combination and im-

plementation is not really integrated and supported by systematic rules, although

this level is the most challenging element of a KM procedural model.

Instead of detailing the lack in implementation guidance of the previous contribu-

tions, North and Papp later tried to shed some light on the first steps of a KM pro-

ject, when they suggested a system of initial situations which precede KM

(North/Papp, 1999) in order to categorize KM project activities: KM can either

start with new IT and Knowledge Management Systems (KMS), with the estab-

lishment of a CKO, with best practice and knowledge exchange, or with top-down

strategic initiatives. Although these situations were intended as paths, the subse-

quent suggestions for activities are quite hypothetical and how to develop them is

not discussed at all. The collection of situations even shows, that the authors want

to start with some measures, which are not derived from a goal – a typical push

oriented concept, which is representative for the approaches before 2001.

20 This question will be further examined in the chapter on second wave concepts. They

produced an own section of process oriented KM.

Visualization of virtual Knowledge Communities 37

It can be observed, that the approaches which intended to propose procedures for

KM implementation increased after 1998. Still, their underlying push oriented per-

spective is resulting in top-down procedures which derive business activities from

a conceptual KM strategy or in out-of-the-box solutions which just have to be im-

plemented in order to end up with successful KM. For example, the three instru-

ments suggested by North (see paragraph above) actually do not derive their ap-

plication field.

The premature and somehow unsatisfactory stage of KM theory development dur-

ing these years can be further substantiated by highlighting some conclusions

drawn by Lehner in 2000 (Lehner, 2000:231), which also summarize the status

quo. There is consensus, that KM is a systematic procedure to identify, collect and

store knowledge in an organization. However, concurrent approaches show that

there is no consensus about how to actually do that. Hence, this missing clarity

about ‘how’ to do KM is highlighting the poor system behind all existing ap-

proaches and their lack of interrelation. Hence, taking the initial maturity model of

KM theory development in Figure 8 on page 26, the solution oriented stages have

been tried to approach by theories but during the first wave, they have not yet sat-

isfactory solved this challenge.

The restrictions of the underlying push orientation of the introduced approaches

can also be implied by the reflexive summary of Lehner (2000:261). At this time,

typical KM project procedures were starting with the definition of the organiza-

tional responsibles and their task profiles21. Taking these specifications as a start-

ing point, the continued with the establishment of the according incentive system,

then set up the IT system, and defined objectives. There is no general preceding

stage of situation analysis, despite first individual analytical concepts like e.g.

from Schueppler (1996).

Economically, this push orientation resulted in the predominance of a rather tech-

nical perspective similar to Information Management (Lehner, 2000:144). This

again was propelled by an increasingly simplified focus on explicit knowledge ar-

tifacts, i.e. documents or contents (thus information) despite the theoretical propo-

sition, that KM requires a comprehensive sociological element because human

knowledge carriers have to be included. This implies a big need of improving cor-

porate applications of information systems and Information Management to re-

main competitive. However, following this development, around 2000, most cor-

porate projects technocratically concentrated on implementing some KM software

and thus did not really changed their way to manage Knowledge Work. Some-

times the budget was only spent for enriching the intranet, without recognizing the

other necessary project domains including change, incentive, human resource,

21 Without telling, how these tasks can be derived. Actually the definition of the role to

carry out tasks is the final step after important tasks have been identified from busi-

ness situations.

38 M.Trier

process, and communication management. This is also substantiated by a survey

of KPMG (2003). The strong dependence on technology led to the undermining of

theories by New Economy software suppliers. They managed to bring the term

KM into close relationship to their software products and thus affected its usage.

This lead to the problem, that big companies increasingly associated KM with

their software products (often equipped with a completely subjective and product-

related implementation method). Being dependent on these small and shallow

software products, which often were far from being a mature application, the dis-

cipline of KM was heavily affected by the economic downturn in 2001, which

eliminated about 90 percent of all ‘KM’ software vendors22. As only few organ-

izational measures for learning and knowledge transfer were in place, the product-

centered KM approaches became affected, too.

In conclusion, the first wave of approaches moved towards pushing KM structures

and technologies into a company using a top-down approach. The theoretical re-

search was not applicable, as the researchers at that time usually only separated

the fields into various segments, which (were and) could not be translated into

practical project-methods or organizational operations and structures (‘concept

giants’ vs. ‘implementation dwarfs’). This irritated companies, which preferred to

apply a simpler but working KM concept by choosing and implementing applica-

ble software packages (e.g. via simply visiting software exhibitions). All corporate

projects concentrated on selecting often very fancy KM platforms or other IT

components relevant for KM. This resulted in too much hype of consultancies and

technology vendors, who only repackaged their methods or products to sell it un-

der the new label KM. Usually this resulted in acceptance barriers, as complexity

increased because there is simply one more application to be used by the work-

force. Mostly, IT responsibles where allocated as project managers. They viewed

KM from their technical perspective and often enough did not check beforehand

the acceptance of the used tools in the various departments. The focus was on

documentation and codification: trying to build huge databases of relevant docu-

ments was hence the major objective of a first wave of KM that was technology

driven and enthusiastic about artificial intelligence.

Knowledge Managers did not employ any strategy or tried to influence corporate

culture towards the motivation of sharing knowledge. They also neglected the

necessary creation of back office structures and reengineering of knowledge in-

tense processes. Often enough the budget was just calculated for the IT implemen-

tation, and hence there was no money left for change management, training, inter-

nal marketing and the like.

22 In an internal KM project conducted in 2003, a software survey resulted in the insight

that out of about fourty vendors which offered what they call KM solutions before

2001, only about four survived to offer their product in 2003. This is by far a more re-

alistic number to serve the required market demand for KM software.

Visualization of virtual Knowledge Communities 39

Finally, the inability to measure investments led to considering expectations as an

investment rule. As money markets prove, this behavior leads to a very unstable

behavior, as expectations can suddenly decline. In corporate KM, the boom im-

mediately declined as soon as the expensive projects were running as the accep-

tance and usage of the provided tools was disappointing. However, there are also

examples (e.g. Siemens, PriceWaterhouseCoopers, or Ernest and Young), where

enough additional effort has been made to augment the tool with various organiza-

tional measures. Moreover, the successful approaches regarded the tools not as

their primary focus but as what they are – tools to support something with IT.

These companies managed to become successfully employing KM to generate

more effectiveness in their divisions.

Immediately after the economical recession of 2001, only little interest for KM

was left in companies. They preferred to deal with their core business issues al-

though this is a very dangerous ignorance given the background of the increasing

role of Knowledge Work. During these times of cost-reduction and down-sizing

(cf. chapter 1), a big advancement for the improvement for the operative imple-

mentation of KM was the idea of applying the business process perspective to

Knowledge Management. This perspective shifted the focus away from technol-

ogy-driven push strategies towards demand-driven pull strategies. It is able to

connect KM to the operations as it proved successful in analyzing knowledge

workers demand and gave a proxy for the efficiency potentials of supporting

knowledge work and knowledge workers. With this trusted co-concept Knowledge

Management moved towards more analytical approaches which are dominant in

the second wave of KM theory evolvement. The people and the actual work situa-

tion and environment moved into the focus.

2.2.3 The second wave of KM concepts – Pull-orientation

Around the year 2001, KM approaches changed their focus towards the analysis of

a company using a KM perspective. This was due to the decreased interest of en-

terprises in trying Knowledge Management methods and the increased pressure to

move away from abstract top-down methods towards an improved, direct and con-

ceivable link to corporate practice and restrictions. In this successive second wave

of KM approaches, which will be analyzed in this chapter, the instruments became

more operative, the projects more focal and the project managers more risk avers

and rational. A core set of methods emerged and can now be implemented into

corporate practice. This correlation to the economical cycle also implies that to a

large extend the field of KM has initially been bound to the somehow immature

self-understanding of the new economy. However, the subsequent mature ap-

proaches should not be ignored just because of this initial stage of adolescence.

The effect may have been fuelled by the insight, that there is no single valid theo-

retical KM solution possible for the multitude of corporate settings. The concepts

40 M.Trier

emanating from this development can be regarded as belonging to a ‘second wave’

of KM theory evolvement.

As this wave is still quite young and not as dispersed as the previous, it can not be

represented well on a timeline. However, the most important contributions are dis-

cussed and listed in Figure 10 on page 52.

Opposite to the technology and top-down oriented push strategy of the first wave,

the approaches in this segment can be characterized as pursuing pull oriented

Knowledge Management which starts by identifying the demands of Knowledge

Workers. Afterwards, they derive KM applications and methods. Here obviously a

thorough method for requirements analysis is fundamental.

To support this endeavor, the corporation as the object under consideration needs

to be modeled from a knowledge perspective. Here, an overview about the actual

KM related entities (which can be analyzed) in a corporation is helpful. A model,

which supposes a very general yet comprehensive such overview is the KM entity

model of Trier (2003, 2005) shown in Figure 9. It assembles all relevant entities

together with their interrelationships on a very aggregated level and can thus serve

as a starting point to analyze a company. The respective model elements are proc-

ess/activity, document, employee, and topic. These elements have been found nec-

essary to analyze in corporate KM projects conducted by the author. Connecting

these four main entities automatically directs the focus to very relevant relation-

ships between the concepts. For example, an author is connected with other au-

thors, with topics, his documents and of course with the processes or activities he

is responsible for.

The KM entity model now allows for allocating different approaches to check for

their coverage of the four domains and their interrelations. For example, the early

KM approaches of the first wave, which treated knowledge as an object, focused

on the entity document with the attached authoring person and the related topics or

other documents. This perspective thus ignores the entity process/activity and

various relations between authors and processes as well as the direct knowledge

exchange between different authors. The limitations in terms of a low coverage of

the model’s KM entities lead to reducing KM to Information Management and

Data, Document, and Content Management.23

A further and very widespread approach to collect requirements for an analytical

methodology for KM is to follow the sequence of business activities of the respec-

tive employees and hence the relation between the entities employee and process

activity. If now the potential application of KM instruments to support important

knowledge objects and topics related to the activities of the business processes is

23 Which is of course a very important aspect (and result) of a KM implementation, but not

its primary objective.

Visualization of virtual Knowledge Communities 41

concerned, this process analysis is called process oriented Knowledge Manage-

ment (also cf. Wargitsch, 1998, who first explicitly demanded this perspective).

Figure 9: The KM Entity Model.

The KM entity model highlights the importance of the business process oriented

approach to KM. It covers the process (as a sequence of activities), the attached

responsible person and the related documents necessary for the transactions. Still,

the domain of topic is missing. Here, recent modeling approaches for knowledge

intensive business processes extended the conventional method in exactly this di-

rection to increase the coverage of the entities in the KM entity model. The main

innovation has been the introduction of topic elements into process models (see

for example Gronau et al., 2003). Still, the numerous interrelations between em-

ployees, connections between employees and topics or topics and documents, or

dependencies between documents are not included. There are two directions pos-

sible to overcome this limitation. First, process orientation can move towards a

multi-perspective approach, which only uses processes as their main orientation

but then also includes document structures and other network views on the KM

entity model. The alternative option is to employ a different primary perspective.

It has to be an analytical perspective which starts from the actual knowledge carri-

ers, the people who do the Knowledge Work, and includes their networks and

various connections to topics, documents, and processes. Whereas the business

process oriented approach assumes codification and defined task structures as the

primary objective, the network oriented approach starts from the employee and is

less deterministic as it assumes that task structures are a changing contextual re-

sult. However, both approaches can be combined.

The network oriented approach to Knowledge Management will now be discussed

and derived in more detail throughout the subsequent sections of this book. For

this people network oriented approach, the existing network oriented approach of

Employee

Process/Activity

Topic

Document

Business Process Management

Communities of Practice

42 M.Trier

Communities of Practice (CoP) constitutes the main analytical perspective and

KM instrument (e.g. Allee, 2000; Enkel et al., 2002; Seufert et al., 1999). As part

4 will show, they actually contain networks of people which are gravitating

around topics (Wenger, 1998) and documents. Returning to the KM entity model

in Figure 9, CoPs comprise a different content-, people-, and network oriented

perspective next to the transactional and process oriented perspective or the ex-

plicit knowledge objects (document) perspective. In this way, CoPs are a valuable

and complementing instrument within Knowledge Management in corporations

(Enkel et al., 2002).

Summarizing, the main contribution of the KM entity model is that KM needs to

provide for a better transparency about the enterprise, by managing not objects but

the multiple relations between entities and moreover between the various instances

of one entity: e.g. which employees are related or which documents are related.

With this approach two major strands of analysis and management can be re-

vealed: next to process oriented KM, a people and network oriented KM approach

is to be applied to cover the complete requirements of corporate Knowledge Man-

agement. This is in line with the two main KM strategies proposed by Hansen et

al. (1999), namely Codification and Personalization (see section 2.2.2). Figure 9

also shows, that the process oriented and the people- or network oriented perspec-

tive are indeed very complementary so that a combination of both approaches

might also be possible. This is thus substantiating the idea of combining Codifica-

tion and Personalization (Hansen et al., 1999) despite Hansen’s hypothesis of mu-

tual exclusiveness in actual corporate applications.

The two perspectives together form the main concepts of the second wave of KM

theory contributions. After the year 2000 the KM discussion moved away from

‘concept giants’ to focus on a more operative level (however, still recognizing the

important strategy and management layer). Both second wave approaches were

increasingly discussed, analyzed and developed to enable a more effective man-

agement of corporate knowledge. The two main strands will now be briefly intro-

duced to complete the longitudinal observation of how the discipline of Knowl-

edge Management evolved.

2.2.3.1 Process oriented KM Approaches

Starting in 2001, the disciplines of Knowledge Management and Business Process

Management merge towards an integrated process oriented Knowledge Manage-

ment approach. Next to the explanatory categorization of the last section, in an al-

ternative explanation approach, which also aims at classifying different research

approaches, Abecker et al. (2002) differentiated three different layers. Both ap-

proaches differ in their perspective. Whereas Trier (2005) proposes the main enti-

ties or domains and identifies differences between approaches by analyzing their

coverage and main perspective of these entities, Abecker et al. try to establish lay-

ers of different granularity in order to position business process-oriented knowl-

Visualization of virtual Knowledge Communities 43

edge. On the top layer, strategic business process oriented Knowledge Manage-

ment is a top-down perspective, which derives knowledge objectives from the

long-term business objectives. The bottom layer deals with KM design based on

communication analysis and diagnosis. It primarily deals with communication as-

pects of knowledge work and develops appropriate methods or tools. It is thus

very hard to be separated from the middle layer, where Abecker allocates ap-

proaches of business process oriented design, where methods and tools for busi-

ness process analysis are extended to meet the new requirements of Knowledge

Management24. This middle layer is dealing with modeling methods derived from

business process management and the modeling of existing processes to find po-

tential for improvement. Here it has to be noted, that although differentiating lay-

ers might look like a feasible approach to segment the various KM approaches,

this concept does not create the impression, that in practice actually every layer

must be instantiated in a meaningful KM approach. Strategy has no use if there is

no connection to processes and communication. This is why in this book, the en-

tity oriented segmentation as presented in the last chapter is preferred. However,

the three levels can also be interpreted as a maturity model, as the one introduced

in Figure 8 and are thus similar to the classification into two waves pursued in this

book (a first wave of strategic analysis and a second wave of a complementary ap-

plication of process and network based KM, similar to Codification and Personal-

ization). With this perspective, network oriented KM appears as the most ad-

vanced and most detailed approach.

The main existing approaches which belong to Abecker’s process category are in-

troduced now, although it has to be added that all lack in specification. The main

contributions to this field are CommonKADS, Promote, BPO-KM (GPO-WM in

German), and KMDL. Additionally, a further approach will be introduced, which

has been developed in practical project work by the author to aid as a project pro-

cedure for process oriented Knowledge Management projects (POKM method).25

The CommonKADS approach (Schreiber et al., 2000:17) is a method which ini-

tially was developed in a big European Research Program to generate a methodi-

cal support for knowledge engineering.26 This established approach to Knowledge

Engineering has been developed long before 2001 but influenced the second wave

KM concepts and their capturing of knowledge intensive business processes to a

large extend. Although its original objective of constructing a program that can

perform a difficult task adequately is completely different, its process of Knowl-

edge Acquisition can be regarded as similar, because Knowledge Acquisition in-

cludes the elicitation, collection, analysis, modeling, and validation of knowledge

24 In central Europe they comprise the predominant part of current KM approaches.

25 An overview about current process-related approaches can also be found in the work of

Remus (2002:37).

26 This related discipline of KM is mainly analyzing a corporate domain to generate knowl-

edge-based systems, like for instance expert systems.

44 M.Trier

for Knowledge Engineering and Knowledge Management projects (Milton, 2003).

The according Knowledge Acquisition (KA) techniques have been developed to

help with the elicitation of knowledge from an expert. Obviously, they focus the

same issue of examining and analyzing knowledge intensive business processes.

However, the special methods for documenting the expertise of a knowledge

worker are not completely applicable in business process oriented KM. This is

supported by Abecker, who comments, that the method contains only a few KM-

specific concepts and focuses technical infrastructures (Abecker et al., 2002). An-

other example for the divergence can be derived using the interview structure for

domain-knowledge elicitation provided by CommonKADS (Schreiber et al.,

2000). It consists of the steps identifying a particular sub-task, ask the expert to

identify “rules” used in this task, take each rule, and ask when it is useful and

when not. This rule based approach gives no special procedure to ask for the top-

ics of expertise and their relation to the process.

A further approach is the PROMOTE method (Hinkelmann et al., 2002), which

integrates strategic planning with the evaluation of Knowledge Management and

Business Process Management. The intended scope of the approach covers the

analysis, the modeling, and the execution of knowledge intensive processes. It ex-

tends the more general method of Business Process Management Systems (BPMS)

including strategic decision, reengineering and resource allocation, workflow and

performance evaluation. The additional KM related steps are creating awareness

for enterprise knowledge, discover knowledge processes, create operational

knowledge processes and organizational memory, and evaluate enterprise knowl-

edge. The second element, discovering knowledge processes, deals with capturing

knowledge intensive business processes. In the underlying reengineering stage of

the BPMS method, process knowledge is getting documented. It consists of the

sequence of activities and its related employees, organizational units, necessary

data, application systems, and resources. The PROMOTE method now extends

this process knowledge to additionally capture functional knowledge, consisting of

the identification of knowledge intensive activities, the description of relevant

knowledge flows, and the identification of knowledge flows between persons and

processes. Although this approach introduces a systematic orientation for the nec-

essary elements to be captured in the context of knowledge intensive business

processes, a detailed description of how to elicit these additional items in a KM

project and a discussion of the actual pitfalls in the practical implementation

would be beneficial.

The next process oriented approach after 2001 is BPO-KM (in German:

GPO-WM®) by Heisig (2003). It proposes a method for a process oriented analy-

sis and design of Knowledge Management Solutions. Within this procedure of

eight steps, the KM Audit analyzes the fundamental conditions including the

evaluation of existing IT systems, the analysis of the information- and knowledge

culture, and the determination of the demand for information and knowledge. The

main focus of this step is the identification of potential for improvement of the ex-

Visualization of virtual Knowledge Communities 45

isting utilization of knowledge in the business context. The subsequent step ana-

lyzes knowledge intensive processes to identify strengths and weaknesses or pos-

sible improvements. Further, the process- and task-related demand for knowledge

is identified. Unfortunately, the procedural model does not propose any methods

which explain how to capture and model processes on a detailed level and hence,

this approach results in a rather strategic analysis of business processes and con-

centrates on the identification of strengths and potentials for the utilization of

knowledge.

A final sample approach of the second wave of KM is currently being developed:

KMDL® (Knowledge Modeling Description Language) and the related tool

K-modeler. It aims at providing a structured procedural model and the necessary

tools to identify and model knowledge intensive business processes. Based on

classical process modeling notations, this model includes topics and their proper-

ties in the generated model. Using the model, different KM measures can be de-

rived (Gronau et al., 2003).

An own process oriented KM (POKM) project approach has also been developed

in corporate projects in the years 2002 to 2004 (Trier and Mueller, 2004). A spe-

cial focus is put on a detailed method for capturing knowledge intense processes

in an interview. It hence it extends the existing broad perspective on this aspect of

KM projects. It allows for insights into questions like how to comprehensively

capture expert processes, how to store the collected information in a structured

way and how to design supporting instruments and materials in order to generate a

high quality data set, which subsequently is utilized to derive Knowledge Man-

agement measures.

These procedural models are just some examples of the multitude of available ap-

proaches. They all provide a comprehensive theoretical model which can help to

plan and execute a process oriented KM project. However, their practical execu-

tion in corporate settings necessitates detailed procedural specifications of how to

actually execute the proposed steps. Further methodical discussions and founda-

tions are required for issues like how to comprehensively capture expert processes,

how to store the collected information in a structured way, and how to design sup-

porting instruments and materials in order to ensure a high quality data set, which

subsequently is utilized in every type of KM project to derive management meas-

ures.

2.2.3.2 Community oriented KM Approaches

As section 2.2.3 already identified, the conventional process oriented approach is

not covering relationships between knowledge workers and relations of knowl-

edge objects to a topic. Rather, it employs the process model as the main connect-

ing location, e.g. as the indirect link that can be found between two similar ex-

perts. The implicit assumption is that they share their process (and work) context

and are hence related. Different topics are arranged along the process. However,

46 M.Trier

there exists a direct relation to the community oriented approach: Knowledge

Work in knowledge intensive business processes consists of information process-

ing activities; these are consisting of information acquisition processes. One main

method to acquire information is to actually ask related people. Over time a con-

tent oriented cooperation and a special yet informational network structure be-

tween related people emerges next to their process oriented business tasks. The

resulting groups are called communities and comprise a complementary view on

knowledge work with a topic or a knowledge carrier oriented perspective27. It al-

lows for deriving and observing networks of topic oriented knowledge-objects (i.e.

topic taxonomies or semantic networks) or networks of experts in topics (i.e. so-

cial networks).28 As shown in section 2.2.3, this second strand is represented by

the research on Communities of Practice and constitutes an instance of the ‘per-

son-to-person’ Knowledge Management strategy, that Hansen et al. named Per-

sonification (Hansen et al., 1999).

The community approach is also implicitly contained in Kuehn’s and Abecker’s

(1998; also see section 2.2.2) concept of the process-centered approach, which

mainly understands KM as a social communication process. It regards knowledge

as something, which is closely tied to the person who developed it and which is

shared mainly through person-to-person contacts. The main role of IT is hence to

help people communicate knowledge, not to document or store it.

The communication-centered community approach (or Personification) augments

the management of information resources management by dealing with the actual

environment in which people can develop and share knowledge.

Hansen et al. (1999) even frame their Codification vs. Personalization dichotomy

in a way, which proves the high value of the community approach for the difficult

issue of sharing tacit knowledge:

“Thus Hansen, drawing on social network analysis has developed a more contingent

view of KM. This argues that the relative emphasis on KM strategies (personalization

or codification) and associated network links (strong or weak) need to vary according

27 A knowledge carrier is carrying knowledge and is hence a person. However, it has to be

noted, that some approaches also regard a document as carrying explicit knowledge.

Still, this practice should not be adopted, as according to epistemology (part 2.1.1) it

only carries information.

28 An alternative but generally consistent differentiation is being supposed by North et al.

(2000). The authors segregate between technocratic, expert oriented, and ecology ori-

ented (networks) Knowledge Management. Whereas the technocratic approach is cur-

rently moving towards process orientation (using the business process as the nucleus

around which knowledge processes happen), the expert oriented approach should

rather include a perspective with multiple experts in networks. North’s approach is

obviously seeing the difference between expertise, which is profiled and made trans-

parent to others in yellow pages and the active connection and communication be-

tween experts in communities, thus forming a social network.

Visualization of virtual Knowledge Communities 47

to (i) the purpose of the task at hand (exploration versus exploitation) and (ii) the kind

of knowledge that is important to achieve it (tacit or explicit).” (Swan, 2001:3)

The community perspective on KM is obviously important for knowledge explora-

tion with a high share of tacit knowledge. Taking this to the extreme, it could even

be argued, that sharing of tacit knowledge is the only true concern of KM as shar-

ing explicit knowledge via codification is actually touching the less complex (but

still required) domain of Information- and Data Management and the results of

such systems however have been disappointing (Roschelle, 1996; Davenport and

Prusak, 1998; cit. in Hildreth et al., 1999:349). Instead, Communities of Practice

should thus be the dominant perspective for approaching a management of knowl-

edge in an enterprise. This is also substantiating the KM Entity Model and the en-

tities to be covered by KM, as introduced in section 2.2.3.

The difficult challenge of transferring tacit knowledge and the role of communi-

ties are also analyzed by other researchers. Hildreth et al. (1999) interpret the the-

ory of Legitimate Peripherical Participation (Lave and Wenger, 1991) as a direct

link between Socialization of Nonaka (as a theoretical option for transferring tacit

knowledge in a company, section 2.2.2) and communities as a means for socializa-

tion in a distributed organizational setting.

“As a first step towards the management of such knowledge we need to understand

the social processes that govern its construction and its sustenance in an organization.

Lave and Wenger (1991) suggest that soft knowledge is created, sustained and shared

through Communities of Practice by a process called legitimate peripheral participa-

tion (LPP). They describe how groups are regenerated by newcomers joining and

eventually, replacing existing members. The newcomers learn from “old-timers”

through co-practice that is graduated, permitting them to undertake more central and

critical tasks. In so doing, they not only learn the domain skills associated with the

practice but they also learn the language of the community, its values and its attitudes.

Through this kind of participation newcomers move from peripheral positions to more

central ones and in so doing are transformed into old-timers. Membership is legiti-

mated though participation and participation is legitimated through membership.”

(Hildreth et al., 1999:350)

Hildreth et al. further relate the Community of Practice to actual knowledge types

it generates:

”We can discern three trajectories of soft knowledge construction in these communi-

ties. Firstly there is the gathering of domain knowledge (for example, how to solve a

particularly tricky diagnosis problem). Secondly, the construction of knowledge of

work practices specific to the community (for example, knowledge of an individual

machine’s idiosyncrasies and how they are catered for). Finally there is the knowledge

that the community constructs about the competencies of its members.” (Hildreth et

al., 1999:350)

A final statement which highlights the importance of a shift towards a people-

centered perspective on Knowledge Management as represented by the Commu-

nity of Practice comes from Swan (2001), whose “research indicates” that:

48 M.Trier

“It seems more likely that the key to achieving coordinated action does not so much

depend on those 'higher-up' collecting more and more knowledge as on those 'lower-

down' finding more and more ways of getting connected and interrelating the knowl-

edge each one has.” (Swan, 2001:8)

Obviously, there is a socially constructed nature of knowledge itself to be found at

the core of Knowledge Management. This is why this book is looking at this per-

spective and trying to find approaches to manage knowledge in an enterprise by

capturing and analyzing its networks of people. However, in order to leave enough

room to thoroughly elaborate the related issues of this book’s main topic, commu-

nity oriented approaches are just briefly mentioned in this section in order to allow

for a complete time line of Knowledge Management theories at the end of chapter

2.2. The theoretic notion of social construction of knowledge is derived and dis-

cussed as a part of this book’s approach of a community and network oriented

Knowledge Management in part 3. All details on community related definitions,

classifications, structures, processes, benefits, gaps etc. are introduced in chapters

4 to 6. Finally, a novel technical solution to aid the community and people-

network perspective and to develop a comprehensive instrument for network ori-

ented KM is described in part 7.

2.2.4 Current Reception of KM in Corporate Implementations

To enrich and complete the picture of how the discipline of KM is constituting and

evolving, this general section will finally introduce some findings from a recent

KM report of KPMG (2003) which reflects the actual corporate application of

Knowledge Management approaches. Simultaneously, this integrates part 1 on

economic developments with the current sections about the respective theories.

Among the respondents of the survey, obviously even during recession (or simply

because of it), Knowledge Management has been considered a strategic asset by

four out of five companies. 78 percent believe that they would even be missing out

on business opportunities by failing to successfully exploit available knowledge. If

asked for quantifying these impressions, companies estimate that, on average, six

percent of revenue as a percentage of annual turnover or budget is being missed

from failing to exploit knowledge effectively.

There is one more positive indicator, the interest of board members. They are the

ones who allocate budgets and bring KM into the divisions. So it is vital to get

them excited about the topic and the survey proves that in half of the companies,

the involvement of the board members increased in the past three years. Although

it just the half, as it may be argued now, it is sufficient a number to make KM a

prospering concept.

The difficulties in measuring returns from KM projects can be implied, when 64

percent of the survey’s respondents say their project’s ROI is unknown. Of the

remainder, 27 percent report ROI above required company level and nine percent

Visualization of virtual Knowledge Communities 49

report ROI at required company level. The results of KM projects indicate where

the improvements come from, that save money. Four out of five Companies use

Knowledge Management to realize synergies among units, two thirds accelerate

innovation or reduce costs, three quarters achieve higher customer added value, 70

percent improve quality, and one quarter is trying to reduce exposure to risks (26

percent). 50 percent of companies report that these improvements resulted in clear

financial benefits and returns. Among the non-financial benefits, companies ex-

perienced quality improvement (73 percent), increased teamwork (68 percent), in-

creased speed and response (64 percent) as well as better decision-making by

frontline workers (55 percent).

Interesting is also the project size or in other words the average KM spending. It is

less than two percent of revenues. It can be inferred, that the cost of KM is rela-

tively low in comparison to the business opportunities it can exploit. In compari-

son with 50 percent that reported financial effects and three quarters, that report

non-financial benefits, KM hence seems to be a useful investment.

Often it is advisable to start with KM in a pilot area. To answer the question of

what division to pick, the business areas which applied KM are of interest. How-

ever, there is no single most relevant area to be found. Rather, KM is widely

spread throughout the company and there is actually no area, where there are no

improvements possible. The core application is found in areas that are close to the

market and the customer like marketing and sales, service delivery, and delivery

with 53 percent, followed by operations with 51 percent. 32 percent operate KM

in the distribution channels and 26 in procurement. In the areas, that are not di-

rectly located at the value chain, i.e. the back office, KM is applied as well. 43

percent use it in functional areas like human resources or R&D and 36 percent in

strategy.

Many insights can be won, if the major barriers are analyzed that were encoun-

tered when introducing a successful Knowledge Management in a company. In 50

percent of the companies KM was not being integrated in the business processes

and subsequently generated problems. Only about 25 percent state, that their KM

system is rarely used, hence, the acceptance of such systems seems to be possible

to achieve. Four out of five companies claim, that KM is not being treated as a

daily priority, but rather as something carried out only in special occasions. This is

also implied by the 60 percent that feel, that not enough time is being allocated to

KM activities. It is a business activity that does cost time in the first moment but

saves it later and that effect is often not being accepted by business managers. In-

teresting is also, that the creation of a knowledge-sharing culture is being seen as a

major obstacle by two thirds of the respondents.

This survey proves again, that a sophisticated strategy and the according tool is

not the primary issue. Rather, the practical barriers to overcome lie in the actual

connection of the expertise of knowledge-sharing groups of people to everyday

business issues.

50 M.Trier

2.3 Resulting Framework of theoretical KM Approaches

After having reviewed the development of two stages of theoretical contributions

to the research discipline of KM and after having introduced the macro-economic

environment and micro-economic corporate adoption of KM methodology, a brief

summary of the resulting framework can be generated.

Taking the maturity model of KM theory evolvement of Figure 8 as a broad guid-

ance, the two waves can be illustrated as shown in Figure 10. The various intro-

duced theories are summarized and the main interrelations are coded as directed

edges. On the left hand side, a broad timescale is added to show the chronological

order. The gap in theory development in the mid nineties can be observed as well

as the two strands of the second wave after the year 2000. The current book adds

two main ideas at the end of the time line – the KM entity model to link process

and community oriented approaches and later, the modeling and analysis of com-

munication networks in order to improve CoP management which in turn furthers

network oriented KM (see chapter 7 and 8).

Finally, to enable the reader to compare this classification approach the very com-

prehensive categorization of Earl (2001) is related to the two waves categoriza-

tion29. This enables to compare the approaches and helps to arrive with a more ob-

jective perspective. In his distinction of different methodological strategies of KM

approaches Earl (2001) identified seven ‘schools’ of thought:

1. "Systems school" which tries to codify knowledge in databases or expert

systems,

2. "Maps school", which tries to map knowledge or the knowers (e.g. yel-

low pages),

3. "Process school", which tries to improve business processes with the

storage and transfer best practices,

4. "Trading school", which tries to manage intellectual property,

5. "Organizational school", which is based on the concepts of social net-

works,

6. "Spatial school", which redesigns the physical work spaces, and

7. "Strategical school", chiefly based on Grants (1996) resource based view

of the firm, which tries to change the strategical view on the firm.

They can be aligned with the first and second wave categories introduced in the

previous chapters: Documentation is the technology-driven approach of the first

29 The classification introduced by Abecker et al. (2002) includes the three stages (strategic,

business process level, communication level) and has already been discussed and re-

lated to the two waves classification in chapter 2.2.3.1.

Visualization of virtual Knowledge Communities 51

wave (Earl even notes, that they did not come far), similar the strategic approaches

belong to the first wave’s concepts and their idea of top-down management, the

Process School is obviously the process oriented categorized as an important

strand within second wave. Intellectual Capital and Spatial School are not dis-

cussed in this book at all, still both comprise parallel concepts. The former aims at

linking corporate accounting with Knowledge Management and the latter includes

architectural design. Finally, the Maps and Organizational school indeed together

form exactly the perspective this books adopts. Although providing a comprehen-

sive classification, this scheme does not convey the various logical and chrono-

logical relations as it is done in the lifecycle oriented two waves approach.

Following this strategy, the next sections will derive, discuss, and explain the in-

creasing role of the network oriented perspective in organizational theory in order

to show that a network oriented Knowledge Management is required to support

such structures of intelligent organizations. Then the support of such a network

oriented KM with IT applications is discussed to derive a concrete solution from

business requirements in chapter 8.

52 M.Trier

Kleinhans 1989:

Technical and

personal KM domain

Albrecht 1993:

Tasks

Pautzke 1989:

Layer Model of the

Org. Knowledge

Base

Wegener 1987:

OM is a transactive

memory

Hedberg 1981:

Organisational

Memory

Walsh/Ungson 1991:

structure of OM

‚Containers‘ for K,

Polanyi 1985:

Tacit and explicit

knowledge

Nonaka 1991:

Tacit and explicit

knowledge flows in

an organization

Schüppel 1996:

Analytical approach 4

steps model for KM

Schneider 1996:

KM: demands

operational

measures

Probst 1997:

Knowledge process

steps: 8 bloick

Schüppel 1997:

combine human

capabilities with

supporting technol.

Davenport 1998:

Succesfull project

elements

Hansen et al. 1999:

Emprically find

codification vs.

personalisation

Nonaka 1991:

Five prerequisites of

organisational

environment

Grant 1996:

Ressource-based

view on KM

Bullinger 1997:

Succesfull project

elements

Various techn.-driven

push-oriented applic.

of KM software in

enterprises

1990

1993

1999

Drucker 1970s:

Knowledge Work

Schreiber 2000:

CommonKADS

Hinkelmann et al.

2002:

Promote

Wenger 1998:

Communities of

Practice

Heisig 2003:

BPO-KM Gronau et al. 2003:

KMDL

People network

or community-

oriented

approaches of

Hildreth et al. 1999:

Properties of CoPs

Cothrel 2000:

Measuring CoPs

Wenger and Snyder

2000:

CoPs vs. groups

Enkel et al. 2002:

Practical CoP

approach

Schoen 2000:

Design of CoPs

Process-

oriented

approaches

of KM

1991

1992

1994

1995

1996

1997

1998

2000

2001

2002

2003

2004

Gongla and Rizzuto

2001:

CoP Lifecycle

1989

1988

1987

1986

<1986

KM Entity Model Modeling and

Analysis of CoP‘s

Communication

Networks

2005

Figure 10: The Time Line of KM Theory Evolvement.

Visualization of virtual Knowledge Communities 53

3 A network oriented Foundation of Knowledge

Management in Organizations - Knowledge

Networks

During the introduction of the development of KM as a research discipline, the

previous chapters already implied the strong connection of Knowledge Manage-

ment and organizational theories. KM is actually conceived as a means of organ-

izational development which aims at the creation and utilization of knowledge in

an enterprise. As already outlined in section 2.2.1, Lehner notes that KM has some

roots in the discipline of Learning Organization and has been motivated by the re-

lated discussion of new organizational forms which accommodate for the in-

creased importance of data, information, knowledge, communication and informa-

tion flows (Lehner, 2000:225; Schneider, 1996). Wiegand (1996) even under-

stands an organization as a knowledge storage facility and Willke proposes that

KM is all about creating an intelligent organization via competence management,

qualification, and the ability to learn (Willke, 1996).

A further clear indicator for KM’s dependence on organizational theories is im-

plied in the extensive definition of Knowledge Management as proposed by Snis

(2001): Knowledge Management is the management of the organization towards

the continuous renewal of the organizational knowledge base. This means e.g.

creation of supportive organizational structures, facilitation of organizational

members, or putting IT-instruments with emphasis on teamwork and diffusion of

knowledge (as e.g. GroupWare) into place. If knowledge itself is regarded as the

capability for effective action, then knowledge is also related to organizational ac-

tivities.

As KM obviously happens in an organization and tries to analyze and influence its

structure, processes and values, it is dependent on the theoretical conceptualization

of the construct ‘organization’ itself. Organizations are thus the object under con-

sideration, which implies that the understanding of organizations determines the

structure of Knowledge Management programs. In other words, it has to be known

what an organization actually is before appropriate Knowledge Management can

be developed in an organization.

In this context, it has to be warned, that an insufficient consideration of the com-

plex underlying organizational mechanisms30 leads to a reduced probability of

success and a lower acceptance of the KM measures which are based on such a

30 This problem is sometimes found in business informatics, which is often reducing or-

ganizations to machines and structures.

54 M.Trier

rudimentary organizational foundation and the related limited understanding.

Here, it can often be recognized, that approaches which conceive Knowledge

Management as Document Management and Information Management employ a

reduced understanding of the organization, in which they are to be embedded in.

To avoid problems resulting from limited underlying organizational theory, this

chapter now briefly reviews and reflects selected organizational theories. Similarly

to the introduction of KM as a discipline, a general chronological approach will be

applied. The underlying hypothesis is that if an increasing role of network orienta-

tion in an organization can be found, then only a network oriented KM can sup-

port the knowledge work in such structures. Starting with the first appearance of

the idea of a social network in an organization, the analysis progresses towards the

discussion of the very important influence of systems sciences before modern

theories about network organization are presented. Finally, the few existing theo-

retical contributions from the field of KM itself are included to identify a collec-

tion of theoretical implications for a foundation of KM, which culminate in a list

of direct requirements for network oriented KM. These theoretically sound re-

quirements can finally help to guide knowledge managers in their creation and

support of an environment in which people can develop and share knowledge.

To anticipate the main conclusion, the following section shows, that KM has pri-

marily to consider the networks of organizational and social actions between em-

ployees in an enterprise or as Sachs puts it:

“If only the organizational (explicit structural) features of work are considered in de-

signing work and the importance of learning is left out, there will be negative conse-

quences in the conception and implementation of [organizational] design”. (Sachs,

1995:38).

3.1 Early Organizational Theories and their Role for KM

A comprehensive historic overview about existing organizational theories has

been provided by Wyssusek (2004:250). He identifies Scientific Management as a

physiological-technical approach (1910-1920, Taylor etc.), further the bureau-

cratically-administrative approach (1920-1930, Weber etc.), the motivation ori-

ented approach (1930-1940, Mayo, Roethlisberger, Dickson, etc.), the decision

oriented approach (1940-1960, Simon, March, Barnart, Kirsch, etc.), the systems

oriented approach (1950-1970, Parsons, Etzioni, Luhmann, Bertalanffy, etc.), and

the interaction oriented approach (1980-2000, Weick etc.). This development in

organizational theories coincides with different assumptions about the role of the

people. Starting with conceiving people as a factor of production and a responsible

for administrative tasks, the perception changed towards accepting motivation, de-

cisions, complexity, or even sense-making as triggers for human behavior in or-

ganizations. Subsequently, the individual act in social group contexts is expressed

by the systems and interaction orientation of modern approaches. This role of the

Visualization of virtual Knowledge Communities 55

individual in the organization is additionally influenced by broad economical and

scientific trends like mass production with low wages, career development, or

automation and Information Technologies (as described in chapter 1).

The previous sections already showed, that even in the first organizational concep-

tualizations (e.g. Taylor, 1911), some interesting implications for the management

of knowledge resulted. For example, Taylor, representing the first group of

physiological-technical approaches in the above classification, observed his

worker’s actions and tacit experiences and captured this knowledge to create job

descriptions from it. This technocratic concept called ‘Scientific Organization’

hence divided labor into small repetitive sequences to organize manufacturing. It

has developed from the famous pin-maker example, which was published by

Adam Smith in the very influential book ‘Wealth of Nations’ about 150 years ear-

lier (Smith, 1767). But interestingly, only a couple of months before Smith’s book

was available, Smith’s teacher Ferguson objected to this origin of the mechanistic

perspective on organizations as machines. By this he almost simultaneously initi-

ated the opposite school of thought, quite contrary to Smith’s first contribution to

economic and organizational theory. Ferguson wrote: ‚Mankind are to be taken in

groups, as they have always subsisted. The history of the individual is but a detail

of the sentiments and thoughts he has entertained in the view of his species: and

every experiment relative to this subject should be made with entire societies, not

with single men’ (Ferguson 1767:I1). With this, Ferguson emphasizes the impor-

tant role of groups and their underlying structures and processes.

Despite Ferguson’s challenging thoughts, the ideas of his pupil Smith and later

those of Taylor about the division of labor and the subsequent mechanistic meta-

phor of an organization as a machine dominated the organizational and economi-

cal theories until the early 20th century. However, Baecker (1999:18) remarks, that

Taylor was progressive in the sense, that he replaced the existing nepotism with a

more rational system. Finally, his approach included aspects of emancipation and

expressed mistrust for the societal traditions.

At that time, the sociologist Durkheim was the first to challenge the dominant

theories by stating that collective life is not emerging from individual life, but on

the contrary, it is the other way around - the latter develops from the first. In the

context of an organization, Durkheim emphasizes, that the division of labor re-

quires a worker to not only concentrate on his specific task, but also to remain in

continuous contact with his neighboring functions to recognize their demands and

changes. This relates to the importance of social embeddedness to improve the

existing division of labor (Durkheim, 1977:320). The sociologist concludes that

the division of labor needs to also create solidarity in order to avoid the break up

of growing social communities in times of increasing competitive pressures to

survive. Here, Durkheim differentiates mechanistic solidarity (social relation-

ships) between similar individuals and organic solidarity between diverse but

complementary individuals. In sophisticated and complex organizations (or socie-

ties), the organic (i.e. complementary) social relationships dominate as the divi-

56 M.Trier

sion of labor and the subsequent specialization of the people is high (Wyssusek,

2004:228). This is one of the first instances where sociological thoughts intermin-

gle with theories of how to organize enterprises. Simultaneously it is a pioneering

inroad into the concept of networked organizations and their according manage-

ment.

In 1960, Mayo reinvigorates this sociological perspective of Durkheim in explain-

ing the results of the famous Hawthorne-Experiments (Mayo, 1960:124). A dis-

cussion of the organizational-sociological role of the Hawthorne-Experiments has

been provided by Etzioni (1967:56-82). Under the direction of researchers of the

Harvard University in the 1920s and 1930s George Elton Mayo conducted a series

of experiments in the Hawthorn-Factory of Western Electric to substantiate a sci-

entific position that is opposing the scientific management movement of Taylor

and which is showing, that social factors in groups are important for a firm’s suc-

cess. In his experiments, he finds, that a person’s “desire to be continuously asso-

ciated in work with fellows is a strong, if not the strongest, human characteristic.

Any disregard of it by management or any ill-devised attempt to defeat this human

impulse leads instantly to some form of defeat for management itself” (Mayo,

1949:111).

Mayo’s colleagues, Roethlisberger and Dickson later built on the results of the

Hawthorne Experiments, when they proposed their management theory of ‘Human

relations’. A main constituent of this was the concept of an Informal Organiza-

tion (Roethlisberger and Dickson, 1947:558):

“Many of the actually existing patterns of human interaction have no representation in

the formal organization at all, and others are inadequately represented by the formal

organization. […] The blueprint plans of a company show the functional relations be-

tween working units, but they do not express the distinctions in social distance,

movement, or equilibrium […] nor does a blueprint plan ordinarily show the primary

groups, that is, those groups enjoying daily face-to-face relations. Logical lines of

horizontal and vertical co-ordination of functions replace the actually existing patterns

of interaction between people in different social places. The formal organization can-

not take account of the sentiments and values residing in the social organization by

means of which individuals or groups of individuals are informally differentiated, or-

dered, and integrated. Individuals in their associations with one another in a factory

build up personal relationships. They form into informal groups, in terms of which

each person achieves a certain position or status. […] Informal social organization ex-

ists in every plant and can be said to be a necessary prerequisite for effective collabo-

ration. Much collaboration exists at an informal level, and it sometimes facilitates the

functioning of the formal organization. On the other hand, sometimes the informal or-

ganization develops in opposition to the formal organization. The important consid-

eration is, therefore, the relation that exists between formal and informal organiza-

tions.” (Roethlisberger and Dickson, 1947:558)

In this context, for Mayo, an important consequence of man’s desire to associate is

the shift from focusing individuals towards focusing groups. Subsequently, means

Visualization of virtual Knowledge Communities 57

have to be developed to enable the management of this new ‘object’ under consid-

eration and which provide insights about the social group as a foundation for

managerial actions. (cf. Wyssusek, 2004:314). This directly supports the notion of

network oriented Knowledge Management in a network oriented organization.

Although social relationships between people defy formalization, and by that also

render measurement, control and management difficult, Mayo beliefs in the oppor-

tunity of a targeted influence on social groups: “The eager human desire for co-

operative activity still persists in the ordinary person and can be utilized by intelli-

gent and straightforward management” (Mayo, 1949:19). This management ap-

proach includes the development of ‚social skills’. By means of this skill, man-

agement should commit itself to the continuous process of studying human situa-

tions – both individual and group – and should run its human affairs in terms of

what it is continually learning about its own organization (Roethlisberger and

Dickson, 1947:604).

These conclusions show, that in the context of the Hawthorne experiments, the

original focus on conventional psychological stimulus-response-mechanisms has

been extended by the observation of influential social factors. Like for example,

Wyssusek (2004:311) shows, this is also proving, that the sole reception of psy-

chological results, as often been found in the subsequent literature, has been an

incomplete influence on organization sciences. Rather, the Hawthorne experi-

ments can be seen as a major breakthrough and milestone for the development of a

sociological perspective and subsequent sociological managerial instruments in

enterprises. This closely relates to instruments to support knowledge exchange di-

rectly between individuals. However, Nonaka and Takeuchi argue, that although

Roethlisberger and Dickson criticized Tayloristic approaches of Scientific Man-

agement and perceived the employee as a social being in a group context, they did

not manage to construct usable theories and hence the Management Theory of

Human Relations was and often still is dominated and consumed by scientific ap-

proaches (Nonaka and Takeuchi, 1997: 49).

From these findings, in the era of decision oriented approaches to organizational

theory, March and Simon developed their comparison between organizations and

organisms: Organizations are assemblages of interacting human beings and they

are the largest assemblages in our society that have anything resembling a central

coordinative system. The high specificity of structure and coordination within or-

ganizations – as contrasted with the diffuse and variable relations among organiza-

tions and among unorganized individuals – marks off the individual organization

as a sociological unit comparable in significance to the individual organism in bi-

ology (March and Simon, 1958:4). This metaphor is another illustrative example

for the development towards a notion of intelligence and knowing in an organiza-

tion.

58 M.Trier

3.2 Shift towards Organizational Sociology

The previous section discussed the fundamental and preliminary shift to the appli-

cation of a sociological perspective on organizations. Taking this view, Scott

(1986:86) defines organizations as social structures, created by individuals with

the goal to conjointly pursue defined objectives.31 He created a summarizing ma-

trix typology which can be utilized to segregate the conventional bureaucratic ap-

proaches (compare Max Weber, 1972) from the emerging systemic perspective.

Figure 11: Typology of organization-sociological Approaches. Source: Scott (1986)

In his analysis of organizational approaches, Scott identifies two differentiating

factors: first the organization can be perceived as a rational or as a natural system

and subsequently it can be segregated between a closed and an open system.

This matrix now allows depicting the shift of the underlying assumptions in organ-

izational theory which has already been implied in the previous section. It is a

shift from the left to the right column. The first approach emphasizes and exam-

ines the structural properties and determinants. Following Scott (1986:45), such a

rational system is an organization that pursues very specific goals with a commu-

nity with very formal and restricted social structures. It requires clear objectives, a

high degree of formalization to standardize the behavior of the organizational

members (elements) and the creation of a structure which is independent of the

individual interests (Scott, 1986). Obviously, the physiological-technical approach

of Taylor’s Scientific Management (Taylor, 1911) and the bureaucratically-

administrative approach by Weber (1972) fall in this category. Among their disad-

vantages are limited responsibility for actions which is replaced by the objective to

perfectly execute atomic tasks without understanding the whole, the application of

rules and control actions for their own sake, and the neglecting of the effectiveness

of informal relationships to solve problems. However, this structure works well in

31 This corresponds with the common German opinion represented by the definition of Kie-

ser and Kubicek (1993:4). They define an organization as a social construct, which is

continuously persuading an objective and is having a formal structure, which is being

formed and aligned to the objective by its members.

Rational systems

(emphasizing structural

dimension)

Natural systems

(emphasizing process

dimension)

Closed systems

(emphasizing internal

dimension)

Open systems

(emphasizing system-

environment relations)

Rationally designed

bureaucracies

Resource- and environm.

dependencies

of organization

Formal structures

are less important

than informal relations

Organization-internal

exchange relations

- persistence

Visualization of virtual Knowledge Communities 59

a stable environment, where people can become specialists and no exceptions dis-

turb the system.

The alternative perspective, which is represented by the second column of Scott’s

matrix, focuses on the dynamic processual behavior in a (closed or open) system.

In this paradigm, members of an organization participate in informally structured

collective activities because of their common interest in the continuing existence

(persistence) of their system. Whereas the rational system’s perspective focuses

the structure of the organization, the foundation of the alternative view is to con-

ceive organizations as consisting of people with much less emphasis on predefin-

ing their structure. This structure is not defined but emerging from longitudinal

power and exchange relationships. This is finally producing an alternative primary

source of advancement, which is also responsible for the approaches’ names:

Whereas rational organizations are defined and developed with the help of a ra-

tional (top-down) plan, the natural organization approach is determined by its

natural (bottom-up) developments, which interact and form an additive movement

or stabilization. Because of these properties, the second alternative is also called

micro-political approach. The organization is a whole consisting of interdependent

interactions or ‘games’ by relatively autonomous individuals (Ortmann et al.,

1990:55). The visible organizational structures emerge as a codification of com-

promises which have been agreed upon in corporate power and exchange relation-

ships – a micro-political construction of formal structures (Ortmann et al.,

1990:68). Obviously, the role of communication and knowledge exchange be-

tween interacting experts is implicitly included. Eventually, the organization is

lead by dominant coalitions pursuing different interests (Pfeffer and Salancik,

1978, cit. in Scott, 1986:156). The natural organization is a further theory which is

emphasizing the importance of autonomous work in groups of people, especially

in times of dynamical environments and disturbances affecting the organization

from the outside. Recapitulating the actual economic situation as introduced in

chapter 1 suggests that these structures become more important in the current eco-

nomic situation.

In 1938, Barnard gave another early definition which follows this strand of expla-

nation of organizations. He views organizations as systems of action, consciously

coordinated by communication, which introduces action, controlled information

processing, and communication as tools for sense making. The main concepts are

hence actions, information processing and communication. Following Barnard, the

main challenge of organizing and organization is the integration of actors with dif-

ferent (complementary) objectives. Again groups are a potent means for this ap-

proach.

On an abstract level, the two general paradigms of organizing – rational versus

natural - have also been compared by Baecker (1999:19). He differentiates into a

well-defined system with pre-determined states (trivial machine) and defined tran-

sitions and into a poorly-defined system with unknown transitions between differ-

ent states which changes over time. If now a person is confronted with the well-

60 M.Trier

defined system with fixed rules that do not change, his limited rationality and his

lack of being challenged will disturb this system: The person will employ all his

limitations including misinterpretations of inputs, ignorance of errors or his im-

manent objective to maintain static stability and there is no such system, which

would work with this simplification over a longer period of time. On the other

hand, if the person is confronted with a poorly-defined system which is hostile,

surprising, and instable, just like a modern organization, he will manage to estab-

lish a role and a position within it. He directly reacts to his related persons: The

system consisting of a poorly-defined system and people with limited rationality is

developing towards a well-defined system. Baecker explains that the well-defined

system did not challenge the person, but people want to be challenged. They only

want to be defined by a system, which they can define in turn (influencing the

structures to respect them). People seem to have unconscious abilities for informa-

tion processing which are only activated by unclear situations. They prefer gossip

to clear information and autonomously select trusty information sources and col-

laborators.

Although these two perspectives of rationalist versus systemic (or natural) organi-

zation are opposites in their paradigm, in practice both do not need to be mutually

excluding each other. Rather they get combined when organizations have areas, in

which formal rules are dysfunctional or even inapplicable for efficient results.

Here, informal rules like role definitions, action patterns and expected behaviors

extend or replace and specify rough formal rules. By that, the organization can

adapt these areas to the changing situation. This implies, that especially fast adapt-

ing areas, e.g. with high potential for innovation or with close contacts to dynamic

competition32, are apt to follow the systemic paradigm or in other words: these

domains need to build their organizational structure on loose and systemic group

work. A further generic but very important example for such areas is knowledge

work with its varying results and difficult formalization.

Looking at these organizational areas in more detail, the systemic paradigm is

leading to change at the ‘borders’ of the system: The direct contact of the work-

force with the customers is not being avoided anymore and the top-management

can restrain from the responsibility to exactly prescribe a-priori what the tasks of

the employees should look like as the base must now decide based on environ-

mental signals. This leads to the competitive competence to allow and manage in-

creased levels of complexity within the company. This in turn requires self-

organization with organizational structures that are able to reflect and react to its

own irritations and incorporate its own history in its actions. Systems Science re-

places the rationality oriented approaches of organizing (Baecker, 1999:12). The

32 Examples may be Research and Development, Pre-Sales, or Service. With increasing

customer orientation, the influence of market dynamics affects ‘deeper’ layers and

segments of an organization.

Visualization of virtual Knowledge Communities 61

only prerequisite is hence, that the company provides the necessary ‘space’ (free-

dom and autonomy) to let such structures grow and optimize.

Deal and Kennedy (1982) express this duality when they form their concept of

secondary jobs. They are no formally defined jobs, but jobs in the relationship

network of an organization, jobs as spies, story-tellers, priests, souffleurs, etc.

This is another very illustrative concept which leads to the notion of simultane-

ously participating in the actual business processes or projects and in a network of

employees which exchanges and utilizes experiences for solving problems and in-

fluencing the organizational actions and developments. Later, in chapter 4, the or-

ganizational form of Communities of Practice will be introduced to serve as an

instance to analyze this mode of work in greater detail.

3.3 Systems Sciences in Organizational Theories

The previous chapter introduced the two opposite paradigms of organizing and

their according employment in different organizational situations. The paradigm

of Systems Sciences has been identified as being applicable for supporting knowl-

edge work in highly dynamic segments of the organizational structure. Their main

theoretical contributions will now be reviewed in more detail to augment the foun-

dation of network oriented Knowledge Management with an abstract systemic

layer.

The analysis of organizations as an open and dynamic social system is the per-

spective of the discipline of systemic thinking. The according strands of theoreti-

cal analysis are Systems Theory (Bertallanffy, 1950), Cybernetics (Wiener,

1948), Information Theory (Shannon und Weaver, 1949) and its application to so-

ciological systems thinking (Parsons, 1937; Luhmann, 1984). The role of systems

sciences for organizations is analyzed by Baecker (1999), who demands the crea-

tion of a terminology which is located at the intersection of Systems Theory and

Organizational Theory. Systems Theory delivers the necessary degree of abstrac-

tion whereas sociology provides the necessary societal relations.

Ludwig von Bertalanffy developed the initial systemic framework with his Gen-

eral Systems Theory (GST). It was built on principles from physics (thermo-

dynamics), biology and engineering. Just like the metaphor of organizations as a

natural system, his theory focuses on complexity and interdependence between the

system’s elements. From the environment, the technical system receives signals

and inputs which are processed and transformed within the system and finally out-

puts are transmitted to the environment again. The transformation processes

within the system are determined by the elements and the relations between the

elements (Grochla, 1970:557). A system thus consists of elements in mutual inter-

action. This tremendously resembles the process of knowing using knowledge and

hence once more suggests the presence of KM elements in organizational theories.

62 M.Trier

The theory’s strong philosophical dimension includes the analysis of the human

mind and society. Here, next to Bertalanffy’s analyses, further pioneering contri-

butions can be attributed to Mead (1934, cit. in Weick, 1995:65) who concluded

that social processes precede individual minds. Similar to the comprehension of

Durkheim, Mead describes mind as a social phenomenon:

„It is absurd to look at the mind simply from the standpoint of the individual human

organism; for, although it has its focus there, it is essentially a social phenomenon;

even its biological functions are primarily social.[…] We must regard mind, then, as

arising and developing within the social process, within the empirical matrix of social

interactions.[…] The evolutionary appearance of mind or intelligence takes place

when the whole social process of experience and behavior is brought within the ex-

perience of any one of the separate individuals implicated therein.” (Mead, 1934)

With these statements, Mead was a predecessor of symbolic interactionism, dis-

cussed by Blumer (1973). Their main object under examination is how individuals

and groups interact in order to create identity through the continuing interaction

with others and how these processes interrelate with objective actions.

Similarly, in his Theory of Social Constructivism, Vygotsky (1930) claims that

knowledge is socially co-constructed, which is a negotiation process by which

shared understanding is reached about a “knowledge object” or knowledge “arti-

fact”. Later in the 1960s, Berger and Luckmann (1966) continued this line of

thought in their concept of the social construction of reality. This substantiates the

notion of knowledge creation in a natural organization. The social settings enable

the Knowledge Workers to co-construct a joint competence by interacting with

each other to share experiences in their topic-oriented expert networks.

Subsequently, Systems Theory was applied for sociology by Parsons and later by

Luhmann. Their contributions can aid as an abstract foundation for the natural or-

ganization, described in the previous chapter. The main elements under considera-

tion are structures, relations and interactions between persons. The elements of

these systems (i.e. persons) have their own values and objectives (Schwaniger,

1995:15). Parsons (1951) examined social systems as systems of social acts. He

hence focused on the unit act and the interaction of the members of social systems

and identified different sub-systems, like the personality system (individual goal

attainment, motivations, desires), the social system (responsible for integrating

individual systems components and influencing the interaction of social actors),

and finally the cultural system (framework of norms).

Looking from systems theory, Luhmann conceives organizations as social sys-

tems, which are solely consisting of communication acts related to each other

(Bednarz, 1988). Hejl, another representative of systems sciences, explains enter-

prises as an organizational form emerging from interaction between people (Hejl,

1984). The major role of communication for an organization is emphasized by

Barnard (Barnard, 1938), who defines organizations as systems constituted by ac-

tions, which are coordinated by communication.

Visualization of virtual Knowledge Communities 63

Sociologically oriented systems scientists thus imply, that a major issue for under-

standing organizations is to understand their communication and their ability to

communicate. Communication is hence one of the key concepts of the 20th century

(Baecker, 1999:22) and the constituting element of social systems. Hence, if an

organization restricts communication patterns it also restricts emerging and adapt-

ing social patterns from taking effect in problem solving procedures in dynamic

environments. Here, a strong link can be found between networked organizations,

communication as their primary means of coordination, and learning as an out-

come of communication. The available theoretical approaches which aim to ex-

plain this relationship and the complexity of communication are discussed now.

This substantiates the understanding that Knowledge Management should not only

document communication, but should utilize the networks of communicating peo-

ple as these include meaningful dialogs which are representing the constructed

knowledge.

The relation between communication as the core concept, learning and knowledge

has been analyzed by von Glasersfeld (1996:48). According to him (also see sec-

tion 2.1.1), knowledge is actively generated and not captured from the environ-

ment because communication is not transmitting meanings and semantics but only

coordinates communication partners. Knowledge thus emerges from experience,

learning and communication between people.

Following this direction, in the sociological and hence organization oriented dis-

cussion of communication, slowly the term ‘communication’ has been liberated

from the engineering comprehension of messages (constituted of packages) trans-

mitted between senders and receivers (Shannon and Weaver, 1949). The main

point is, that the processing of sense and meaning of the messages does not hap-

pen in the communication itself, but in the ‘processing units for meaning’ - the

heads - of people who participate in the communication (Baecker, 1999:53).

Communication thus does not only happen in the channel but in the heads. It de-

velops towards a means for self-referential organization in a group of actors with

mutual perception and hence becomes an ‘emergent’ phenomenon, which cannot

be easily explained due to its complexity. It develops collective argumentation

processes, by which individual knowledge is shared in an organization (Probst and

Buechel, 1994:21, cit. in Henschel, 2000:16). Communication leads to structures

and rules for further communication. These rules are changing over time, form un-

consciously and are hard to explicate and codify (Henschel, 2000:16).

Communication is hence no transmission, but - just as the early empirists and later

systems scientists and constructivists propose - construction. The receiver of sig-

nals has many options for understanding what has been sent and hence enjoys a

high degree of freedom. This ambiguity of communication is due to its dual char-

acter. It contains elements of informing and influencing (‘Information’ and ‘Mit-

teilung’, Luhmann, 1984). The information aspect wants to transport the mere fact

(e.g. ‘this is an apple’), whereas the influencing aspect includes the context. It

contains hints like the following: It has been said now, it could have been said

64 M.Trier

something else, the sender wants to imply something, or the message can be inter-

preted by the sender and the receiver. The receiver can chose if he prefers to inter-

pret the implication or the mere fact. Hence communication is not just simple in-

formation of someone. This is an important aspect to consider: Allowing transmis-

sion and storage of information is not KM as the sender and the context are ex-

cluded and thus much value of the communication itself is lost.

Especially in dynamic domains of an organization, communication is more effec-

tive than information. Here, Baecker (1999:23) argues that the special intelligence

of people in their interaction with poorly-defined systems is their intelligence to

utilize communication (to add value). Analyzing an organization thus prerequisites

the understanding of its communication. This is an important implication for es-

tablishing Knowledge Management in a company: supporting people networks in

natural organized systems has to focus the analysis of the people’s communica-

tion. This major implication is therefore focused in the chapters 7 and 8.

A further outcome of communication is discussed in the systemic communication

theory of Watzlawick et al. (1985, discussed in Henschel, 2000:16). It assumes

cyclic relationships of causes and effects. Every action invokes a reaction that is

affecting the actor again. Watzlawick concludes that a connection between per-

sons is formed by iterative communication relations and can either be influenced

by purposefully defining communication rules (rationalist organization approach)

or can emerge from continuing interaction. From these insights, Watzlawick et al.

(1985:56) derive a concept of communication similar to the approach of Luhmann.

He frames the second axiom of communication theory: Every communication has

a content-aspect and a relationship-aspect, where the latter determines the former

and thus constitutes a meta-communication. Thus, next to Luhmann’s notion of

influence (‘Mitteilung’) Watzlawick puts the relationship-building aspect of com-

munication. This is establishing a link between communication in networks and

the social relationships between people discussed by the discipline of sociology.

Schulz von Thuns (1981) finally integrates these two perspectives in his four as-

pects (or layers) of communication: He proposes the content-related aspect, the

relationship-aspect between sender and receiver, the appell to stimulate behavior

(i.e. action) of the receiver, and a self-explanation (‘Selbstoffenbarung’) about the

wishes and motivations of the sender. From this concept Winograd and Flores

(1987:78) form a classification of communication types, including:

• conversation for action, which targets at triggering action,

• conversation for possibility, which is preceding conversation for action

and deals with detailed questions of the intended action,

• conversation for clarification also precedes the action and identifies the

objective of conversation for action, and

Visualization of virtual Knowledge Communities 65

• conversation for orientation that aims at creating a shared reference

framework, which is necessary to collaborate within a social system.

Henschel (2000:19) summarizes the analysis of communication in organizations

by stating, that it is not the individual which is the subject of an intervention, but

the structure which is emerging from the communication between individuals. Or-

ganizational design and development has thus primarily to be concerned with

communication processes within a social system. This comprehension of Knowl-

edge Work in organizations by management of network structures (thereby elevat-

ing the network to become a management object in its own right) will provide the

theoretical foundation for the support of groups of networked experts. It will be

introduced in the subsequent chapters of this book.

Next to the important link between communication and knowledge, the connection

between knowledge and learning is another important aspect, where systems sci-

ences provides valuable contributions for a theoretical foundation for understand-

ing Knowledge Work in a networked organization. Here, Baecker (1999:11) ar-

gues that systems refer to complexity, complexity refers to self-organization, and

self-organization refers to reflection. This is why systems sciences relate and pre-

pare for organizational learning theories. In a systemic view on an enterprise, the

reflection about actions, experiences and the according organization are providing

the prerequisite for learning processes. Following this connection, in the beginning

of the 1980s the introduced classic works from Vygotsky, Mead, or Berger and

Luckmann were resulting in a general and recognizable change in learning theory:

a shift from a capturing and storing hypothesis towards the hypothesis of practical

participation, which is embedding knowledge in practical action (cf. Sfard, 1998).

For example, Lave and Wenger (1991) employed an according learning theory in

their work, which defines learning as process of social construction. The knowl-

edge remains in the contexts (and heads), where it has been generated. For the au-

thors, learning, thinking and knowledge are constituted by relationships between

people, which are acting in, with, and through a socially and culturally structured

world (Lave, 1997: 67).

Brown, Collins und Duguid (1989) come to the conclusion, that knowledge and

action are reciprocal categories: knowledge is always situational knowledge and

is being generated by action. This means, that the situation is fundamental for all

cognitive activity: knowledge is related to an action and not to an object, knowl-

edge is related to a context and is not abstract, knowledge is generated in interac-

tion between the individual and its environment, it is neither objectively defined

nor subjectively generated. It emerges as a functional element of interaction and

not a truth (cf. Bereiter, 1994). Again this conflicts with the predominant idea of

current KM initiatives to explicate and store experiences in knowledge bases.

Hutchins (1996) finds that knowledge is not simply a matter of an individual’s

mental representations, but is frequently distributed among the abilities of group

members and the artifacts that they use. Accordingly, knowledge is co-constructed

66 M.Trier

by interactions among people and their shared artifacts, including prominently by

means of negotiation practices that result in establishing a common ground for un-

derstanding (also cf. Vygotzky, 1930; see above).

General Systems Theory, its application to sociology, the analysis of the human

mind in a society and the related conception of learning as social interaction even-

tually lead to the application of systemic thinking to organizations and organiza-

tional theory. The main branches are the theory of Organizational Learning and

the Learning Organization as well as Systemic Management approaches. The no-

tion of learning is again hinting the strong relationship to the management of

knowledge.

The discipline of Organizational Learning applies the notion of feedback proc-

esses and communication relations to examine individual and interindividual or

organizational processes for learning (often simply considered as adaptation) in

organizations. Such learning is a characteristic of an adaptive organization. It

senses changes in its environment and adjusts. Accordingly, this includes learning

from experience and implementation of individual experiences into organizational

learning cycles. Here, Argyris and Schoen (1978:20) distinguish between two

modes of learning with an increasing degree of reflexiveness: single-loop and

double-loop learning. Single-loop learning includes the observation of the differ-

ence between expected and obtained outcomes. This difference is used by indi-

viduals, groups or organizations to modify their actions without affecting the un-

derlying values. Double-loop learning includes the reflection about underlying

values and assumptions that lead to the obtained action and their subsequent modi-

fication. This is comparable to a change in the existing routines to create new

ways and opportunities for decisions and actions. March and Olson (1975) link

individual and organizational learning with individual beliefs in a feedback loop.

Individual beliefs lead to individual action, which is then generating organiza-

tional action (for example following the micro-political mechanisms, introduced

above). This organizational action can cause a response from the environment,

which is processed by the individual and may affect and improve its individual

beliefs and subsequently its action over time. However, organizational learning is

not simply the sum of individual learning’s, as individuals that learn can also be

employed in organizations which are not learning. The model of Kim (1993) inte-

grates Argyris with March and Olson and focuses the analysis of the underlying

information flows in the organizational learning loop in order to detect interrup-

tions: an individual action could for instance be rejected by the organization for

political or other reasons. Related to this field of Organizational Learning is the

concept of the Learning Organization. The latter applies the theoretical findings of

organizational learning to propose normative recommendations about how to cre-

ate organizations that continuously and effectively learn. An example is Senge’s

approach of the five disciplines of a Learning Organization (Senge, 1990). They

include Personal Mastery, which is the individuals’ creative approach to their per-

sonal development. A second discipline is mental models, which highlights that

Visualization of virtual Knowledge Communities 67

organizations are affected by underlying assumptions of its individuals. Shared

visions comprise the third element and suggests, that the integration in enterprises

are not only ensured by hard coordination mechanisms like orders but also by

shared visions, assumptions and objectives. Senge further proposes the develop-

ment of team learning to extend individual learning. Finally, as the fifth required

discipline for organizations which are able to learn and adapt in a dynamic envi-

ronment, Senge proposes system thinking including the analysis of cause-and-

effect chains and feedback-loops.

Such approaches towards Organizational Learning demand self-responsibility and

rich coordination and communication between employees, teams and project

groups. This is consistent with the requirements demanded by systems theory. Ex-

isting approaches to solve problems are continuously discussed, challenged and

improved and require a high amount of flexibility in organizational structures.

However, theories about Learning Organizations and Organizational Learning see

learning as adaptation. That is why they did not yet produce a theory about

knowledge creation (Nonaka and Takeuchi, 1995:61). They face the dilemma of

simultaneously requiring stability or efficiency through routinization and flexibil-

ity, innovativeness and readiness for change.

Next to Learning Organizations, another strategy to apply General Systems The-

ory to organizations is the systemic management theory (e.g. Schwaninger,

1995:4). Its main assumptions include that organizations as open systems ex-

change material, energy and information with their environment. Further, they are

complex, interconnected and dynamic entities, they are self-referential33 and op-

erationally closed (i.e. can not be determined by the environment). The mutual re-

lations of the elements are determined by social and natural rules and comprise the

formal and informal structure of the system (Kirsch, 1969:665). The degree of

complexity of the system is determined by the various relationships between the

elements (Luhmann, 1984:37). The border is being generated by a difference in

complexity, which is being reduced within the system (Luhmann, 1984:37). The

development of sub-systems can cause an increase in the system’s complexity and

the steering and regulation is keeping the system in a dynamic equilibrium and

under control.

Systemic management theories also integrated ideas of cybernetics. Cybernetics is

a further branch of systems science, which has been developed from mathematics

and communication technology. It analyzes the steering of systems and looks for

universal rules and properties which are working like a regulation mechanism for

33 The concept of self-reference assumes that the development and adaptation of properties

and behaviour in systems (the achievement of an equilibrium state) is dynamically

produced by the system’s elements and the operations between them. It cannot be de-

termined by the environment and depends on the pre-defined adaptability of the ele-

ments and their structure (Luhmann, 1984:25).

68 M.Trier

a system to keep it in some equilibrium (Ulrich et al., 1976:138). Major insights of

this technological perspective have been applied to organizations in the discipline

of management cybernetics. For example, Stafford Beer has applied cybernetic

insights about the function of the human central nervous system to organizations

and discovered a structural similarity between living systems (organisms) and or-

ganizations. Further, Burrell und Morgan (1979) analyzed the metaphor of an or-

ganism to explain organizations. They examined general principles of open sys-

tems in an organization, like for example system borders, processes, inputs, out-

puts, homeostasis, subsystems or mutual dependency.

In such a systemic analysis the focus is hence on the observation of the relations

and interactions between elements rather than the analysis of the elements and

their properties. Linear causal relationships are replaced by feedback loops (Hen-

schel, 2000:15). Thus, incorporated complexity, connectedness and dynamics are

necessary aspects to consider for modern organizational management.

Summarizing, systems sciences concluded that social processes precede the indi-

vidual minds, which means, that knowledge is socially co-constructed. This alone

shows the inadequacy of treating knowledge as a storable object, which can be

transferred between people. Rather knowledge emerges from communication and

learning in a group. It can even be distributed over a group. Communication is ac-

tually the means by which knowledge networks are created. Systems Sciences also

refer to learning processes in organizations. If knowledge is socially co-

constructed and learning is the means to generate knowledge, then learning is so-

cial, too. This leads to the distinction of individual and organizational learning

processes which are interrelated and comprise feedback loops. Finally, theories of

Organizational Learning also demand for rich means of coordination between

largely self-responsible people and a high demand of flexibility in the organiza-

tion. In such loosely structured parts of an organization, learning is a way of

adapting the corporation to the requirements of its environments. With this behav-

ior learning and subsequent knowledge work require a corporate environment

which is similar to an organism. It can be concluded again, that KM has to provide

methods for exactly such an environment and thus has to support communication

networks of people in their problem-solving processes.

3.4 Networks, Coordination and recent Organizational

Theories

Whereas systems sciences discuss the terms system, knowledge, learning, and

communication together with their interrelations, their conclusions only implicitly

support the idea of choosing networks as an organizational means. A reason for

this is that Systems Sciences conceive themselves as an abstract theory, which is

not directly discussing the set up of organizational structures.

Visualization of virtual Knowledge Communities 69

Based on ideas of organizational learning via rich and effective corporate commu-

nication, recent theories apply systems science to directly discuss the properties of

networks. This strand of discussion more directly supports the argumentation, why

networks of experts are a suitable perspective for KM activities.

Market Hierarchy Network

Normative

Base

Contracts, Own-

ership

Employment con-

tracts

Complementary

strengths

Communication

Means

Prices Routines Relationships

Method to

overcome con-

flict

Bargaining, Law

Suits

Administrative

Order and Control

Mutuality, Repu-

tation

Flexibility High Low Medium

Commitment

between parties

Low Medium to High Medium to High

Atmosphere,

Climate

Accuracy, Mis-

trust

Formal, Bureau-

cratic

‘open-ended’

mutual advan-

tages

Preferences of

Actors, Deci-

sions

Independent Dependent Interdependent

Mixed forms,

adaptations

Repeating trans-

actions, Hierar-

chical contracts

Profit Centers,

Market-rules, In-

ternal Prices

Formal Rules

Figure 12: Markets versus Hierarchies. Source: Powell (1996:221)

Starting from the assumption that the spectrum of available positions is reaching

from markets on the one end to bureaucratic hierarchies on the other, the question

is now, which organizational structure fits the needs of modern organizational ap-

proaches best. Whereas markets have a very high flexibility as prices help to con-

nect unknown parties, hierarchies work with routinuous, pre-defined, and inflexi-

ble exchanges between fixed partners. In markets, the commitment between the

parties is lower as in hierarchies. Between the extreme positions of markets on the

one hand and hierarchies on the other, the network structure can be located. Its

communication is based on mutual relationships and the reputation of the mem-

bers of the network. The flexibility is much higher than in hierarchies but still

lower as in markets.

70 M.Trier

In market transactions the exchange is specified in detail, trust is not necessary

and contractual commitments are supported by legal rights. In the contrary, net-

work-based exchanges have undefined sequential transactions in the context of a

common pattern of interaction. Sanctions are more normative than legal; the un-

derlying relation replaces the goods (Kowol and Krohn, 1995).

However, it has also to be considered that in practice the borders between the

three paradigms often dissolve as mixed forms exist. A network can establish for-

mal rules or hierarchies, a hierarchy can implement market mechanisms (e.g. in

profit centers) and markets can work with repeated transactions. This is a very im-

portant implication for managing networks, as there is some degree of freedom in

the adoption of an appropriate amount of formalization. The comparison between

markets and networks is summarized in Figure 12.

The previous chapter on Systems Science already identified the important role of

communication for effective organizational structures in dynamic environments.

This communication is a means of coordination between organizational members.

The mode (or the means) of coordination is simultaneously the main differentiator

to segment the available structural paradigms for organizing. Next to the division

of labor and its segregation of general tasks into work packages, coordination is

the second important basic principle of all organizations (Kieser and Kubicek,

1992:95). There are five coordination mechanisms: mutual coordination (informal

communication), personal order (hierarchical coordination with formal authority),

standardization and automation of work processes (programs), standardization of

work products and outputs (Plans), and standardization of required qualifications

of the employees. Mintzberg relates these forms of coordination to complexity

(Mintzberg, 1992:13): With increasing complexity of the work processes, the pre-

ferred coordination mechanism shifts from mutual interactive coordination to hi-

erarchical commands and then to standardization (first of work processes, then of

work outputs and then of qualifications), but only to finally shift back to mutual

and interactive coordination. This sequence substantiates the link between the dy-

namic environment, the resulting increase in organizational complexity and the

shift to interactions in networks of experts.

The same result has been provided by Baecker (1999:10), who argues that the in-

creased level of inherent complexity within organizations should not be perceived

as a problem, but as the solution. Complexity is the requirement to allow the sys-

tem to make errors, detect them, and learn from them (Baecker, 1999:34). This in

turn requires redundancies, variety and even barriers which divide clusters in the

system to stop external disturbances from affecting the whole structure. If the en-

vironment is not observed and ignored in the present, the reality will cause the

problems later (Baecker, 1999:35). Baecker further continues (1999:36), that in-

creased levels of complexity can be accommodated by moving from bureaucratic

structures to communication and hence to network structures. He is thus in line

with the findings of Mintzberg.

Visualization of virtual Knowledge Communities 71

The two authors’ contributions produce a very interesting result, which substanti-

ates, that Knowledge Work in a network is more capable of handling problems,

than Knowledge Work in a pre-defined (hierarchical or process oriented) business

setting with fixed transaction processes.

Another theory which derives an appropriate form of coordination and control

from different organizational settings is proposed by Ouchi (1977). He establishes

the two dimensions measurability of output and asymmetry of knowledge and in-

formation. If the output is measurable and the information asymmetry is high, a

market-like coordination mechanism based on prices is suitable. If the information

asymmetry is low but the output is difficult to measure then hierarchical orders

and process control should be applied. However, if the output is difficult to meas-

ure and the information asymmetry is high, Ouchi suggests mutual and interac-

tive self-coordination with its immanent social control (cf. Figure 13). This consti-

tutes another argument for adopting a networked organization for Knowledge

Workers. Their output is difficult to measure and they generate benefits because

they complement each other. This also implies the likelihood of a high degree of

information asymmetry between the different combined fields of expertise.

Information and Knowledge Asymmetry

low high

Output difficult

to measure

Instruction and Control of

Execution (Processes)

Self-regulation with social

control, self-control

Output easily

measurable

Instruction and Control of

Results

Control of Outputs, Prices

Figure 13: Selecting the appropriate Form of Coordination. Source: Ouchi (1977)

Here, the special organizational form of a network enables the flow of information

and influence from the top to the bottom and vice versa, but also horizontally

through the relations (Stamps and Lipnack, 2000). Barley (1996) comes to a simi-

lar conclusion: If in an organizational structure, knowledge and capabilities were

generated domain-specific, the work is less determined by the vertical chain of

command (influence) but rather by (lateral or) horizontal communication and

collaboration between different groups.

The network delivers fix points for the combination of order and chaos in the or-

ganization. Every employee must be enabled to purposefully switch between the

processing of expected and unexpected information, the necessary orientation is

provided by a network, which is offering active connections (for coordination and

communication), but even more importantly connections, which potentially can be

activated on demand in special situations (Baecker, 1999:26).

72 M.Trier

A further substantiation for the important role of informal mutual interactive

coordination as the usual means for generating solutions and decisions in corpo-

rate practice has been developed by Cohen et al. (1972:2). In the ‘Garbage Can

Model’ the authors state that an organization is a collection of decisions, which

look for problems, further of topics and emotions which look for situations to be

applied to, of solutions which are looking for questions, for which they can be the

answer, and finally of people working as deciders who look for work. In other

words, a set of problems meets a set of solutions in situations where a set of peo-

ple needs to make decisions under time constraints. It is a common aspect of mod-

ern organizations which are characterized by limited rationality (as it is impossible

to utilize all available information for one decision), further by a variety of indi-

vidual objectives among the individuals and coalitions, and by an orientation to

past experiences and rules which are consulted prior to estimating future results.

The occurring situations lead to feasible but not necessarily optimal decisions for

that decisive situation. This model shows that the time and the constellation of

meetings between people are affecting the decisions of a company. Thus the or-

ganization should support opportunities for the right people to meet e.g. by elec-

tronic means of communication. The organizational structure is actually regulating

this combination of people and their access to decision processes and thus the

quality of the decisions.

The network structure for organizations is also being supported by the emergence

of a new technological paradigm, centered on information and communication

technologies (Castells, 2000). The network paradigm even extends to the whole

economy, which is according to Castells characterized by the fact, that its capacity

of generating knowledge and processing information determines the productivity

and competitiveness of economic units. Another indicator is that the network ap-

pears as a new form of economic organization. This can be a network of firms or

segments of firms. Large corporations internally de-centralize as networks; small

companies connect themselves using networks.

A related development which was also fueled by the growing ability to represent,

execute and control business processes via electronic means is the approach of

Virtual Organizations. Such virtual organizations can exist within and between

organizations. Intra-organizational virtual organizations imply the absence of

static structures. Such an enterprise has no hierarchy, no organization chart and no

divisions with specified task descriptions. One main assumption is that modern

computer and telecommunications networks reduce the costs of coordination in a

way, which is allowing firms to achieve production benefits without incurring the

higher transaction costs (Malone et al., 1987).

However, the virtual organization as described in these approaches is a special

situation, where the complete enterprise can exist without hierarchical coordina-

tion structures. Still, the notion of a virtual organization emphasizes the aspect of

adding a heterarchic structure to current organizational architectures in order to

Visualization of virtual Knowledge Communities 73

handle specific situations of coordination and leads to the assumption, that it can

be established in parallel to the conventional organizational structure.

As the most recent theoretical debate, Postmodernist assumptions substantiate the

importance of the network principle for modern organizational structures: Boje

and Dennehy (1993) describe the constituting principles of a postmodernist or-

ganization as work teams of multi-skilled workers, working in flexible networks

with permeable boundaries and flat hierarchies. Efficiency decreases with spe-

cialization, formalization, routinization, fragmentation, and division of labor.

Postmodernist organizations employ a high degree of people-centered leadership

and decentralized control.

3.5 Organizational Network Structures

After having reviewed organizational theories and their contribution to the ques-

tion if and how networks can be a suitable form of organizing Knowledge Work,

this chapter will now have a closer look at the constituents of a network.

In order to approach this special organizational structure, it has to be defined first:

Teubner defines a network as a decentrally regulated system of collaboration be-

tween autonomous actors. It should be a lose form of cooperation, which should

neither only consist of arbitrary interaction nor should it own the density and sta-

bility of cooperation in a formal organization (Teubner, 1992). A social network is

hence an informal, person oriented, trust-based, reciprocal, exclusive interaction

relationship which is lasting over time between heterogeneous autonomous, inter-

dependent actors, who voluntarily cooperate to conjointly achieve a surplus and

for that reason integrate their plans and actions (Weyer et al., 1997:64). Such net-

works are social systems, in which acting persons are bound by relationships. By

their interactions they constitute the network structures, which are in turn the basis

for their future actions.

The classification of networks and their definitions help to assemble the theoreti-

cal properties of network structures in organizations. They include indefinite, se-

quential transactions within the context of a general pattern of interaction. They

rest on normative sanctions as the parties of a network pursue their own interests

at the expense of others (Powell, 1996). The chances for opportunism are re-

stricted by social norms. Networks have a high degree of openness, transcendence,

and ambiguity. Such networks help to assess tacit knowledge within and outside

the organization. Powell (1996:225) argues that it is hardly possible to put a price

tag on qualitative issues like innovation and experiments, know-how or special

approaches to production. These aspects of organizational activity can neither be

dealt in a market nor communicated in a conventional corporate hierarchy. Net-

works are resting on social relationships, trust and reputation, which can be a

cheaper control mechanism as supervision by authorities. These issues are dis-

cussed in more detail in the section 5.3.3.

74 M.Trier

Empirical research on social networks confirms the importance of networks in or-

ganizations from a very different perspective: It has been found, that interactions

of people captured over time form a network structure via communication

(Brass, 1985; Krackhardt, 1991). On a more detailed level of analysis, this branch

of research also found out, that a network is usually not homogeneous, but em-

ploys some structural properties. There are clusters and structural holes. The con-

figurations of the actors’ relations differ and so do their roles in the network. For

example, an actor who connects several clusters can benefit from his function as a

bridge or broker. Further, some ties between actors are stronger than others. Such

strong ties relate to increased loyalty and trust due to the ongoing informal interac-

tion between the authors. On the other hand, strong ties can also prevent the actor

to establish new ties for example due to social pressures and lock-in effects. The

individual properties of actors within networks also relate to power structures.

Some prominent actors attract much attention from others (i.e. are often asked)

and thereby are influential for the development of the whole network. Others are

simply very active and thus influence the network development. These properties

which relate to network structures are very important for the analysis and evalua-

tion of people networks. Chapter 7 will hence introduce a broad variety of such

micro-level network properties in order to develop a system for evaluating and

comparing individual network structures.

In summary the previous section showed, that modern approaches of organiza-

tional theory emphasize the systemic perspective. Here, organizations have to be

observed as a system consisting of elements and relationships. In a complex or-

ganization with a multitude of competing objectives and coalitions, the organiza-

tion cannot be treated as something rational anymore. Instead, limited rationality

in decisions should be accepted. Sociological approaches help to focus the com-

plex coordination constituted by interaction and communication between indi-

viduals. The network as a structural paradigm emerges in the society and as an ef-

ficient means for organizing (appropriate domains in) organizations. Webber sum-

marizes this shift with a strong emphasis on its scope:

“The most universal challenge that we face is the transition from seeing our human

institutions as machines to seeing them as embodiments of nature. Even on a large

scale, nature doesn't change things mechanically: You don't just pull out the old and

replace it with the new. Something new grows, and it eventually supplants the old.”

(Webber, 1999:178)

Similar to the relations between the root sciences of KM, the following figure

shows a more detailed summarizing overview about the connections between the

multiple introduced contributions of organizational science.

Visualization of virtual Knowledge Communities 75

Figure 14: Overview about systemic Sciences and their application to Organization.

As outlined in the beginning of chapter 3, Knowledge Management within an or-

ganization is simultaneously dependent on the organizational properties and af-

fects them with its measures. Therefore, KM should regard the various require-

ments put forward by this complex comprehension of modern organizational the-

ory and move from a management of document collections towards supporting

networks of communicating people.

3.6 Implications for organization and network oriented KM

Next to the requirements identified from the analysis of organizational ap-

proaches, there exists a body of literature which directly discusses how these or-

ganizational requirements should translate into KM concepts.

For example, Bonifacio and Camussone argue that a KM system has to be inher-

ently social:

“A KM system cannot aim only at making explicit the implicit knowledge of indi-

viduals, but rather at facilitating the creation and social dissemination of practical

knowledge. This happens by encouraging the creation and vitality of the Web of in-

formal relationships, which feeds the system of Communities of Practice also through

the use of open, weakly structured and mainly collaborative technologies (like group-

ware systems), capable of supporting the creation of traditional, or even Virtual Com-

munities.” (Bonifacio et al., 2004)

Bertalanffy 1950:

GST

(Biology, Thermo-

Dynamics)

Wiener 1948:

Cybernetics

(Mathematics,

Engineering)

Durkheim (1920s),

1977:

Individuals framed by

social processes

Mead 1934:

mind arising form social

processes

Blumer 1973:

Symbolic interactionism

Vygotsky 1930:

Social constructivism

Berger Luckmann 1960:

Social construction of

reality

Parsons 1951:

Social Systems Theory:

social acts

(Sociology)

Luhmann 1984:

Social Systems

consisting of

communication acts

(Sociology)

Weick 1995:

The process of

organizing

(Organizational

sociology)

Lave and Wenger 1991

Learning is social

(Learning theory)

Brown, Collings Duguid

1989

Knowledge is

situational (bound to

action) (Learning,KM)

Hutchins 1996

Knowledge is

distributed

(Learning theory)

Argyris and Schoen,

1996

Two levels of ind. (and

org.) learning (Learning

theory)

March/Olson 1975

Ind. and org. learning

build a feedback loop

(Org. Learning theory)

Kim 1993

Analytical framework ,

information flows for

Ind. and org. learning

(Org. Learning theory)

Senge 1990

normative framework to

foster org. learning

(Org. Learning theory)

Schwaninger 1995:

Systemic management

theory

(systems science)

Burrell and Morgan,

1979

Organism and

Organizations

(systems science)

Glasersfeld 1996

Radical Constructiviism

Hedberg 1981

Organisational Memory

Wegner 1987

Transactive Memory

Systems

Maturana/Varela, 1984:

sociology and biology

rest on same principles

76 M.Trier

Thus the authors support the identified requirements and propose an extension of

the objective by a more subjective approach to KM. Knowledge should no longer

be treated as something general and abstract (like it is done in the codification

strategy). Rather the social and contextual nature of knowledge shows, that knowl-

edge is a phenomenon which manifests in social relationships. The authors regard

knowledge thus as a practice. This implies, that a knowledge manager needs to be

concerned about supporting business processes by documenting knowledge repre-

senting artifacts, like documents or contents and simultaneously needs to manage

social learning processes in communities. Thus the social learning process and

the artifacts it creates as knowledge objects need to be managed. This position is

substantiating the argument put forward by the KM entity model in chapter 2.2.3,

namely to integrate process and network (people- or community-) oriented

Knowledge Management approaches.

The underlying cognitive approach of the KM theory of Bonifacio and Camussone

refers to the constructivist theories introduced in the previous chapters: Informa-

tion is interpreted (by applying previous experiences and beliefs) and not simply

received. The emphasis is shifting from uncertain knowledge to ambiguous

knowledge.

This requirement for KM to include a systemic perspective and the according phi-

losophical principles is also discussed by Wyssusek et al. (2001) in their approach

called socio-pragmatic Knowledge Management. The authors criticize the com-

mon idea of knowledge being ‘transmitted’ via Information Technology:

“The occasionally assumed transmission of knowledge by information systems is only

a derived phenomenon which results from the fact that some humans “coincidentally”

already have acquired the interpretation patterns they need in order to interpret signs

supplied by information systems in the appropriate manner.“ (Wyssusek et al., 2001)

Following the authors, it is not adequate for KM to focus on the relation between

object and individual alone. Rather, “individual action can only be adequately de-

scribed if its embeddedness in shared processes of setting goals and purposes is

considered as well. Reasoning about individual action must therefore regard it as

being derived from common action with its norms and cultural specifics”. This is

applying the arguments of Durkheim to the field of Knowledge Management. Fur-

ther the authors relate to the construction of knowledge in a social setting:

“Since we regard the processes of cognition as constituted by communities, it is our

judgment that a theory of knowledge must be supplied which takes this fundamental

prerequisite into consideration. […] The primary goal of knowledge management is

no longer a mere transmission of knowledge, but the facilitation of reconstruction of

knowledge within the affected workforce. […] Already within the scope of the con-

ception of knowledge management in an organization and the planning of relevant

methods, community plays a central role.” (Wyssusek et al., 2001)

A similar opinion was given by Nassehi (2000). He builds on the central role of

communication for knowledge creation and suggests not regarding individuals or

Visualization of virtual Knowledge Communities 77

their minds as the carrier of knowledge, but the social dynamics of communication

relations. With its self-referential sequence, it can generate cognitive operations,

which so far have only been attributed to humans. Nassehi refers to Luhmann who

defines knowledge as a cognitive stylization of communication with a tendency

for dynamic changes by learning. Similarly, Swan et al. (2000) argue that KM is

about connecting people with people and people with information to enable col-

laboration and community networking. This finally also clarifies the role of IT in

KM. Knowledge Management practice has to focus the human side and IT is just

the enabler for this endeavor.

As a conclusion of this chapter’s extensive discussion of network-related organ-

izational theories, the following list summarizes the main requirements for the ac-

cording network oriented approach of Knowledge Management:

1. Establish a network structure for collaboration, especially in areas with a

high degree of change.

2. Establish a KM approach that is focusing on communication.

3. Carefully configure the extent of formalization and foster social control.

4. Give room for self-organization and adaptation of the employees.

5. Support problem-related communication of people in networks.

6. Support learning processes as social co-construction of knowledge in so-

cial groups and support the translation of insights into individual knowl-

edge.

7. Support communication in a network by providing multiple means of lat-

eral communication for the experts, which can be activated on demand.

8. Consider the situative aspect of knowledge, which is related to action.

9. Incorporate history in new actions and foster double loop learning.

10. Let people carry out parallel ‘jobs’ in the relationship network of the or-

ganization.

11. Enable the creation of mutual relationships between complementing ex-

perts.

12. Analyze and evaluate the emerging network structure between experts to

improve interventions.

These twelve aspects are important for the framing of a network-oriented ap-

proach for KM and the underlying support with IT.

Summarizing, the network oriented approach to Knowledge Management can be

defined as the extension of the document-based support of knowledge intensive

business processes to systematically support a further and very important mode of

knowledge work: the formation of direct and indirect informal and personal con-

78 M.Trier

tact networks, which utilize communication to connect knowledge workers in

topic-domains and thus develop a valuable source for their productivity.

This approach considers the requirements of modern intelligent organizational ar-

chitectures to cope with highly complex and dynamic environments. However,

this perspective can then be integrated with the other documentation and process

oriented perspectives (also see section 5.1 on the integration of communities in the

organization).

Visualization of virtual Knowledge Communities 79

4 CoP as an Instance of a Network Organization and

an Instrument of network oriented KM

As the previous section showed, corporate organizations consist of a mixture of

formalized structures and rules which exist to achieve the corporate objectives and

a more informal network-like organizational structure, which directly connects

related individuals and thus expresses the sociological aspect of corporate organi-

zations. The constituting element of such an organization as a social system is

continuous and recursive communication between the system’s elements.

The organization can be segregated into various sub-structures (which constitute

sub-systems). The official sub-systems as employed by management are organiza-

tional units, work teams, or project teams. Following the currently predominant

rationalist approach to organizing an enterprise, they are usually very formalized

in their tasks, scope, and membership.

The last chapter discussed the shift of attention towards sociological approaches of

organizing in order to deal with current challenges like fostering knowledge shar-

ing, enabling organizational learning from practice, improving problem solving, or

managing complexity. One result of this change in emphasis is the increasing dis-

cussion and corporate application of an ever-existing but still underemployed or-

ganizational form, which is targeted exactly at meeting these requirements – the

Community of Practice (Brown and Duguid, 1991; Steward and Brown, 1996;

Wenger, 1998; Wenger and Snyder, 2000). Such Communities of Practice can be

regarded as an organizational form, as they are also classified as a social group,

which is self-referentially emerging from interaction and communication. This

form is virtual as often it is not leaving its informal status and subsequently its

borders are hard to observe (Henschel, 2000) due to its ‘invisibility’ (Steward and

Brown, 1996:2).

The concept of Communities of Practice is not new: in medieval centuries, com-

munities of merchants and craftsmen developed and achieved official status as

guilds. They already had an economic and a social aspect. Their members ex-

changed news and experiences, defined norms and together formed a social foun-

dation for the economic and individual development of the participants.

After the importance of systematically utilizing human knowledge has been rec-

ognized and the discipline of Knowledge Management emerged, the potential of

this special social configuration which is deeply embedded in a work context has

been re-discovered (Snyder, 1999:9).

Here, a community is playing an important role for the generation, the transfer and

the application of tacit knowledge. It follows the paradigm of socially constructed

80 M.Trier

knowledge, which for transfer requires personal networks in which it is dissemi-

nated slowly via repeated social interactions (cf. Wyssusek et al., 2001; Swan,

2001). With its self-organizing information logistics, this structure is more suitable

for supporting information processes in a dynamic environment as the pre-defined

and limited communication flows of the formal organizational structure. Therefore

communities require special attention as a complementing instrument for Knowl-

edge Management.

The focus is moving away from systematically storing various information objects

(e.g. by using taxonomies and information retrieval technologies) towards dy-

namic and problem-centered recombined patterns of relationships between indi-

vidual employees with their tacit knowledge and its practical application to spe-

cific business problems. The employee connects his experience with information

from the social connections within his people network to a knowledge, which he

can utilize to solve his concrete problem. By repeating such a selection of com-

munication partners as information sources, a topic oriented group of people is

forming and connecting: Knowledge Work is thus accompanied by the participa-

tion in an informally developing knowledge oriented community. In exchange he

contributes to his colleagues’ problem solving quests. Such a pattern of direct

people-to-people connections is representing the personalization approach, first

introduced by Hansen et al. (1999).

This Community of Practice therefore represents the appropriate concept to dis-

cuss the support of knowledge-driven employee networks, which are vital for

modern organizational forms. They can thus be perceived as an ‘instance’ of a so-

cial organization and due to its special advantages can qualify as a suitable and

even central instrument for corporate Knowledge Management.

4.1 Definitions and Basic Properties of Communities of

Practice

A first clarification of the organizational phenomenon ‘Communities of Practice’

can be achieved by analyzing the various approaches to define the term.

First works in analyzing the properties of communities came from the area of so-

ciology. For example Hillery (1955) demanded that a community requires a locus

(shared geographical location) and interaction. Although the requirement for a co-

location has later been omitted, interaction is a constituting factor. However, it is

not necessary that members communicate actively, rather there has to be the op-

portunity to communicate. This again requires the existence of a communication

channel and the consciousness about the existence and the belonginess to the

community.

Visualization of virtual Knowledge Communities 81

In the 1990s several authors worked on the application of this special group form

to enterprise needs. Their work resulted in a detailed discussion of the main prop-

erties of Communities of Practices.

A first definition of Brown and Gray (1995) already highlights the limited group

size, the relationship developed from continuous collaboration, a common pur-

pose, and the access to each other’s knowledge. The groups do also often remain

informal and even invisible for the majority of organizational members (also com-

pare for Steward and Brown, 1996:2).

“At the simplest level, they are a small group of people who’ve worked together over

a period of time. Not a team, not a task force, not necessarily an authorized or identi-

fied group. What holds them together is a common sense of purpose and a real need to

know what each other knows.” (Brown and Gray, 1995)

Brown and Gray also find that over time communities also develop a shared un-

derstanding as a shared cognitive base (Brown and Gray, 1995:81). The definition

of Snyder in 1997 also emphasizes the aspects of this relation to learning and of

informality which together directly refer to the social learning theory, introduced

above.

“Communities of Practice consist of people who are informally as well as contextu-

ally bound by a shared interest in learning and applying a common practice.” (Snyder,

1997)

The relation to knowledge creation is discussed again by Brown and Duguid in

1998.

“Collective practice leads to collective knowledge, shared sense-making, and distrib-

uted understanding that doesn’t reduce to the content of individual heads.” (Brown

and Duguid, 1998:96)

In 1998, Etienne Wenger, who is often referred to as the thought leader and the

pioneer of the business application of Communities of Practice, contributed his

definition, which is often cited, though it actually names no new aspects and is

very similar to the one of Brown and Gray, three years earlier.

„A Community of Practice defines itself along three dimensions: its joint enterprise as

understood and continually renegotiated by its members, the relationships of mutual

engagement that bind members together into a social entity, the shared repertoire of

communal resources (routines, sensibilities, artifacts, vocabulary, styles, etc.) that

members have developed over time.” (Wenger, 1998:2)

The notion of a continuously negotiated focus shows that there are no pre-

definitions of tasks but rather complex dynamics which advance a community.

Wenger further adds, that CoPs can also transcend organizational boundaries

(Wenger 1998) and later develop properties like a strong sense for community and

intensive cooperation (Wenger and Snyder, 2000: 139).

82 M.Trier

Another proponent of a corporate application of Communities of Practice is

McDermott. He stresses the mutual help for problem solving and the aspect of al-

igning the member’s practices and ideas into a common approach to a topic field.

This shows that its members share interests or activities.

„Communities of Practice are groups of people who share ideas and insights, help

each other solve problems and develop a common practice or approach to the field.”

(McDermott, 1999)

A final refined definition of Communities of Practice by Wenger (Wenger et al.,

2002) which is describing the corporate embeddedness of a CoP defines it as a

group of people, bound to the organization as an informal structural unit with vol-

untary and reporting-independent participation of members who share a concern

and a passion about a knowledge domain, care collectively about this domain

(stewarding) and apply the shared practice in their business processes.

Most definitions of Communities of Practice employ the term group. This term

can be further specified.

Sociology regards a group as a social system, which is constituted by communica-

tion processes (e.g. Schein, 1980:146). It is thus similar to the understanding of

the concept organization as introduced in the previous section.

Gebert and von Rosenstiel (1992:122) define a social group as comprised of direct

interaction of members over a long period of time, shared values, norms, and ob-

jectives, the perception of being a part of a social whole, and the formation of

roles within a group.

It can be differentiated into small groups of up to five members and large groups

of 20 to 25 members (Schein, 1980:146). However such a restrictive border is

problematic as factors like the assigned tasks, qualification etc. are influencing

factors. Further, as will be described later, the choice of communication medium

strongly affects the appropriate or possible group size. Another dichotomy is the

differentiation of primary (emotionally bound, close, e.g. family) and secondary

groups (more rational, pursuing tasks) (Staehle, 1991:243). Finally formal and in-

formal groups can be differentiated. The former usually stem from formal organ-

izational structures to achieve a defined goal (like a persistent division or a tempo-

rary project team) and employ a power hierarchy. The informal groups are sponta-

neously formed by ongoing relationships and contacts. This differentiation is often

not easily possible (Staehle, 1991:246) as both types usually employ some formal

and informal aspects at once. Informal groups often develop informal leaders.

A final concept of group research, which is important for Communities of Prac-

tice, was presented by Tuckman (1965:384). He segregated four phases of group

development, which are also to be found in the development of CoPs: forming,

storming, norming, and performing.

Visualization of virtual Knowledge Communities 83

The main constituting properties of CoPs are thus (also compare for Nickols,

2000):

• a strong sense of identity tied to the community,

• the practice itself is not fully captured in formal procedures, and

• people learn and become seen as competent in concert with others.

• Continuing mutual relationships – harmonious or conflicting (i.e., regu-

lar, work-related interactions, rough or smooth).

• shared ways of doing things together, e.g. common practices,

• a rapid flow of information between and among members,

• quick diffusion of innovation among members,

• conversations which come quickly to the point,

• problems which are quickly framed due to a common understanding of

the milieu in which they all operate,

• broad consensus among the members about who is "in" and who is "out",

• shared awareness of each others’ competencies, strengths, shortcomings

and contributions,

• common tools, methods, techniques and artifacts such as forms, job aids,

etc. ,

• common stories, legends, lore, "inside" jokes, etc., and

• evolving shared language (e.g., special terms, jargon, "shortcuts" such as

acronyms, etc.).

Similar to this list, Hildreth et al. (1999) assembled a visual overview about the

main properties of Communities of Practice (also cf. Figure 15). Central are the

properties of community, situated learning, dynamic evolution, social interactions,

participation, communication, a common purpose, a common background experi-

ence and a common language. Possible additional characteristics are informality,

success stories, similar working areas, and voluntary participation.

84 M.Trier

Figure 15: Central Elements of Communities of Practice. Source: Hildreth et al. (1999)

The introduced properties of Communities of Practice have been applied by Hil-

dreth et al. (1999) to establish a method, which can support the identification of

CoPs in an enterprise. In their questionnaire aiming at the identification of CoPs in

a company with 1500 employees, they analyzed if the following five properties

could be found:

1. Are you in regular contact with colleagues/peers doing similar jobs in

other locations (‘similar jobs’ in distributed locations)?

2. Do you talk with colleagues/peers in other locations when they have a

problem to solve (problem solving)?

3. Do you share projects with colleagues in other locations (common pur-

pose)?

4. Do you swap anecdotes/experiences with colleagues in other locations

(narration)?

5. Do you learn from discussions with colleagues in other locations (com-

mon language, narration and Legitimate Peripheral Participation; also cf.

Lave and Wenger, 1991)?

These properties can finally be summarized in a general definition: A Community

of Practice is an invisible and informal structure with a limited size and emerging

from voluntary participation and interaction which over time establishes working

and social relationships. Its informality and invisibility invoke complex group dy-

namics which are focused around knowledge, expertise and learning for solving

problems in a central topic domain which over time integrates, forms, maintains,

and develops a shared cognitive base and shared resources.

Visualization of virtual Knowledge Communities 85

4.2 Typologies of Communities

The term community can be broken down by classifying special forms with spe-

cial properties. However, it is difficult to create a clear separation or to assign real

CoPs to one of the theoretical categories. Still, this also implies the dynamic and

transcendent character of this organizational structure. Nevertheless, the attempts

to build categories can help to outline several specific properties which help to

further the understanding of CoPs.

A widespread differentiation is made between Communities of Interest (CoI) and

Communities of Practice (CoP). The main differentiator is the notion of develop-

ment towards an objective. Here, a CoI is a group of people with shared interests

that does not develop knowledge (e.g. problem solving methods) to achieve a joint

competence in a defined subject field. As the name already implies, usually this

special form emerges in open internet communities which bring together special

interest groups, like for example for gardening or home improvement issues, soft-

ware users, or collectors. However, they also share their expertise in a topic do-

main and hence there is no clear separation possible (cf. Wenger, 1998). Figure 17

implicitly includes a comparison of the properties of both forms, as proposed by

Wenger and Snyder (2000), as their Informal Network34 is a type of a Community

of Interest.

A different terminology but the same differentiation is made by Reinmann-

Rothmaier (2000). She separates innovation oriented from communication ori-

ented communities, which again implies the orientation towards a general objec-

tive - or in other words: the progress - as the difference. However, Reinmann-

Rothmaier concludes, that innovation also prerequisites communication and thus

this classification is only employable for illustration. A solution for this problem

could be to simply detect the notion of innovation in the individual community or

the evolvement of expert clusters.

A further distinction is suggested by Wenger et al. (2002). They classify helping

communities, best-practice communities, knowledge-stewarding communities, and

innovation communities. Helping communities are focused on community rela-

tionships to help each other solve everyday problems, e.g. by posting requests for

help in a threaded discussion and respond to other people’s questions. One exam-

ple is Schlumberger which connects its scientists and engineers. Best-practice

communities are focused on practice, develop, validate and disseminate the ac-

cording methods and practices - for example engineers and operators at Ford Mo-

tor Company, who describe new practices and assess effectiveness and benefits.

Knowledge-stewarding communities are focused on explicit knowledge and or-

ganize, upgrade and distribute knowledge used every-day. One example is Cap

34 However, as a Community of Practice is also an informal network, this is why this very

general term is not used outside the figure.

86 M.Trier

Gemini Ernst & Young and its communities, which focus on finding, organizing

and distributing documents according to topics. Innovation communities are rather

focused on tacit knowledge to foster unexpected ideas and innovations. One ex-

ample is DaimlerChrysler and its communities which encourage engineers to as-

sess new directions in research and which provides a channel for innovation.

Again, this distinction cannot be utilized to separate actual instances of CoPs as a

knowledge-stewarding community shares best practices to help each other and

thereby supports innovation (e.g. as refining best practices).

By looking at the members and their affiliations, a final classification identifies

three categories: business-to-business communities, employee-to-employee com-

munities, and business-to-customer communities (Cothrel, 2000:2). Although the

term employee-to-employee community is describing the same as the more com-

mon term Communities of Practice, introduced above, the term business-to-

business community emphasizes the special constellation when members come

from different autonomous organizations, which usually collaborate alongside the

supply chain. The communities are employed as a single point of information. An

example for such an integrative approach is the joint business-to-business com-

munity ‘Covisint’ which has been initiated by General Motors, Ford Motor Com-

pany, Renault/Nissan, and Daimler-Chrysler. Members are purchasers, sellers,

constructors, engineers, and suppliers. The platform fosters the interaction among

these people from different companies with tools for online-collaborative devel-

opment (Covisint, 2002). Another implication is the application of moderated

communities to maintain and improve customer interest and customer relations

with business-to-customer communities. They are also often referred to as cus-

tomer community or consumer community. The customers can exchange informa-

tion about products (like for example a software application or a software devel-

opment environment for a programming language). The enterprise hosts the com-

munity platform and learns about the customer’s problems, interests and opinions

to subsequently incorporate this knowledge in market strategy and product devel-

opment.

B2B B2C E2E

Members Customers,

suppliers,

distributors, etc.

Customers Employees

Objectives Stronger

relationships

Insight

Efficiency

Innovation

Revenues

Stronger

relationships

Insight

Lower customer

acquisition costs

Revenues

Stronger

relationships

Insight

Efficiency

Innovation

Revenues

Figure 16: Community segments of Cothrel (2000).

Visualization of virtual Knowledge Communities 87

A further special type of community is the Learning Community. It employs pro-

fessional lecturers or external consultants to qualify a group of learners. This can

be done in class-rooms or in E-Learning courses. The actual Learning Community

is emerging during the learning process, after the students achieved an atmosphere

of trust and mutuality to help each other. Usually this special type of community

ends after the courses are over (short life-span) but various contacts between the

group’s members remain and improve the participant’s personal informal net-

works within the company (cf. Trier, 2002).

Finally, a very important distinction can be made between designed versus grass-

roots communities. The latter are also called bottom-up or emerging CoPs (cf.

Schoen, 2000:118). The difference is, whether distributed Communities of Prac-

tice have evolved from initial informal contact between members or from an offi-

cial imposed grouping (also cf. Hildreth et al., 1999).

This last property is very important for the employment of CoPs in an enterprise.

The selected degree of formalism already starts with the initial creation of the

community (bottom-up by members or top-down by management). It largely af-

fects the opportunities for management intervention. This aspect is discussed in

the chapter 5.8 on management tasks throughout the active life of a CoP. Further,

the aspect of designed structure is also the main differentiator in the comparison of

CoPs with other forms of group work in an enterprise. This comparison is required

to fully understand the special configuration of this mode of group work. This is

why this issue will now be discussed in more detail.

4.3 Community of Practice vs. other organizational forms

Despite the many properties which are necessary to call a group a Community of

Practice it is still difficult to define, how big the differences to the existing organi-

zation need to be to constitute an informal community entity. Here, North et al.

(2000) note, that otherwise everything in a company could be segregated into

overlapping community groups as all these groups have a joint learning and work-

ing process with a shared vocabulary and methodology. Another difficulty is it to

combine or compare it with other informal organizations and communities. This is

why it is necessary to further differentiate Communities of Practice from other or-

ganizational forms like formal organizational units (i.e. business division) or pro-

jects, but even more from existing working teams.

Some of the differences were already implied in the definitions like the aspect of

membership (or composition): In a CoP, members get invited. Their participation

is mostly voluntary and self-selected, whereas in a formal organization member-

ship is predefined by hierarchical reporting procedures. Similarly, project team

members are also pre-defined by some senior management person.

88 M.Trier

A CoP’s purpose is it to exchange knowledge and to develop the members’ capa-

bilities; it is not directly aimed at accomplishing a business task like in a project

team or at delivering a product or service like in a formal organization. Hence, its

general independence of concrete tasks is a special property.

Further dimensions are bonding (cohesiveness) and duration, which help to differ-

entiate CoPs from project teams or formal work groups. An important aspect is the

major role of a common identity and mutual commitment in a CoP – this identity

is as cohesive as the achieving of project goals in a project team or organizational

goals in a formal work group. Further the duration of a community is as long as

the topic is interesting for its members. This makes it an interesting alternative for

project members which have finished their work in a project, but are still inter-

ested in the general topic. They can connect to CoPs to keep updated in their field

of expertise and to contribute their valuable insights to other business issues.

Communi-

ties of Prac-

tice

Formal

Work Group Project

Team Informal

Network

Purpose

To develop

members’ ca-

pabilities; to

build and ex-

change

knowledge

To deliver a

product or

service

To accom-

plish a spe-

cific task

To collect

and pass on

business in-

formation

Compo-

sition

Members

who select

themselves

Everyone

who reports

to the group’s

manager

Employees

assigned by

senior man-

agement

Friends and

business ac-

quaintances

Bonding

Passion,

commitment,

and identifi-

cation with

the group’s

expertise

Job require-

ments and

common

goals

The project’s

milestones

and goals

Mutual needs

Duration

As long as

there is inter-

est in main-

taining the

group

Until the next

reorganiza-

tion

Until the pro-

ject has been

completed

As long as

people have a

reason to

connect

Figure 17: Comparing CoPs and other organizational Forms.

Source: Wenger and Snyder (2000)

Visualization of virtual Knowledge Communities 89

Next to a comparison between CoPs, formal workgroups and project teams, the

literature discusses the difference between teams in general (not only for project

work, where it is tied to a specific assignment, after which the group disbands) and

Communities of Practice (cf. Storck and Hill, 2000).

If a team is not working in a project, it is assigned to a specific function or proc-

ess. Often, it consists of very heterogeneous roles which together are solving

tasks. These tasks often change. A team is usually smaller than a community. The

degree and frequency of interaction in a team is on average higher than in a com-

munity. Teams coordinate by shared goals, communities have shared fields of in-

terest. Teams are mostly admistratively defined and their roles are determined.

Hildreth et al. (1999) call this the legitimizing process. They claim that in a team,

legitimizing occurs by role assignment, whereas in a CoP legitimacy is established

through interaction about its practice. They work more often on shared artifacts

than communities where the exchange of personal artifacts is higher. A team does

more often report on its output and activities. Accordingly, there are more control

mechanisms. Quite contrary, a Community of Practice employs very little control

mechanisms. It usually defines its own agenda, objectives and duration (Wenger

and Snyder, 2000). Hence, it can also be formed spontaneously. It redefines itself

according to the needs of its members, which also implies that its purpose is tar-

geted at the member’s interests and needs (Storck and Hill, 2000). The following

table shows an alternative and detailed comparison of teams and CoPs:

Teams Communities of Practice

Team relationships are established

when the organization assigns peo-

ple to be team members.

Community relationships are formed

around practice.

Authority relationships within the

team are organizationally deter-

mined.

Authority relationships in a Commu-

nity of Practice emerge through in-

teraction around expertise.

Teams have goals, which are often

established by people not on the

team.

Communities are only responsible to

their members.

Teams rely on work and reporting

processes that are organizationally

defined.

Communities develop their own

processes.

Figure 18: Comparing Teams and CoPs. Source: Storck and Hill (2000)

As already indicated in the introduction of the term Virtual Community, these

classifications show different archetypes of organizational forms. However, in

practice there are no such clear borders. For example, in a Virtual Community,

physical communication can also play a major role (e.g. to create trust). Similarly,

90 M.Trier

a Community of Practice can develop towards a team or a task force: It could es-

tablish some semiformal system of rules, roles, and hierarchies (e.g. newcomers

versus seniors) and could define the achievement of a specific result or objective

as its primary goal. However, the main differentiating factor is not the absence of

structure but the origin of its definition. It is specified from the community itself.

This is also implying the complexity of the issue of coordinating, steering, or

managing such groups without to overly disturb their underlying group dynamics.

This issue will be discussed in great detail in the next chapter.

Visualization of virtual Knowledge Communities 91

5 Organization and Coordination in CoPs

After having established a distinction between the Community of Practice and

other organizational forms, the actual relation of the community to the rest of the

organizational structure is of interest. Although it is neither a team nor a formal

organizational unit, a CoP has relationships to these other forms and usually

members of communities are simultaneously associated to one or more of these

organizational structures, too. This results in a synergy in subject fields but a com-

petition for work time, which again raises the issue of measurement and evalua-

tion. Costs and benefits of simultaneously belonging to multiple groups have to be

estimated in order to ensure an efficient allocation of every employee’s time for a

systematic employment of such a multi-dimensional structure.

The following sections will first discuss the issue of how to connect communities

to the rest of the organization. Afterwards, practical approaches to install commu-

nities in enterprises are introduced to give an impression of their organization. In

corporate applications, the mixture of supporting emerging communities and the

purposeful definition of such groups leads to a complex management issue. Al-

though communities can inherently not be managed by conventional means, en-

terprises want to influence their development and their work. The resulting ques-

tion of how to actually support and coordinate Communities of Practice in enter-

prises also relates to the issue of changing dynamics and changing properties as

Communities of Practice are subject to a lifecycle which is related to their matur-

ity.

5.1 Integrating Communities into existing Organizational

Structures

The integration of recognizable communities into an organization is a major chal-

lenge. There are various and complex dependencies to consider. A thorough un-

derstanding of the connection between a CoP and the organization is necessary.

A theoretical approach to conceptualize the relation between the organizational

structure and a community is the Hypertext Organization of Nonaka and Takeu-

chi (1997). Its primary concept is the parallel co-existence of three different but

related contexts within an organization: On the central layer of the business sys-

tem, the operative transactions are conducted. It can either employ a bureaucratic

functional or a process oriented organization. A further element of the model is the

project team layer, where multiple project teams are developing new knowledge,

for example in new product development projects. Finally, there is an underlying

knowledge base layer. It is responsible for recategorizing existing knowledge and

92 M.Trier

putting it into new contexts. This layer is no instantiated organizational unit but it

emerges from the conglomerate consisting of knowledge carriers (experts) and

means for documentation or storage of knowledge. All three layers co-exist in an

organization and comprise a dynamic feedback loop: As soon as projects are fin-

ished, people distribute their achieved knowledge in workshops and documents to

infuse the business layer. They can as well personally move back to the operative

work. From this concept, Henschel (2000) develops the implication that organiza-

tion members are simultaneously assigned to a business unit, a project, and the

community.

Nonaka’s concept implies that Communities and the formal process or project or-

ganization are two parallel domains in the enterprise, which have to be combined

efficiently. This issue has been discussed by North et al. (2000). The authors see

the arising challenges in the desire to legitimize the Knowledge Community. Such

a legitimization is usually connected to the access of resources, i.e. time for par-

ticipating in the communities or financial resources for experiments and imple-

mentations. Further it is related to the expectation to produce results. Here, an-

other challenge for the integration of communities in organizations is that the

group is utilized or instrumentalized by the organization. Enterprises could be

tempted to assign projects like restructuring or preparing decisions. In this case,

such an intervention is likely to change the community into a task force. The ques-

tion is, whether this foregoes the initial internal impulses which let the community

emerge or whether the community continues to exist next to the task force.

Being involved in ‘two worlds’, members can perceive conflicts between their

loyalty for the Knowledge Community and the loyalty for their division. Here, the

communities sometimes create an autonomy, which fuels solutions that have no

chance for implementation. This can lead to insights or efforts, which can not be

applied to the ‘normal’ value generation processes in the business units.

Another very important relation between a CoP and an organization is that if the

enterprise provides required time and budget, this generates a demand for control-

ling and monitoring. Here, it is vital, that results and progress of Knowledge

Communities are actively reflected, visualized and communicated. This is not lim-

ited to proving the generation of financial values, but can include qualitative im-

provements, time savings, spread of best-practices, etc. This last issue is one of the

main motivations of this book. In the end, a software tool is developed, which

supports the generation of transparency in a company and the communication of

value generation or progress by visualizing the networking of people and their

communicative work.

Practical examples for approaches towards a successful integration and for the de-

velopment towards an actual feedback circle of Organizational Learning can be

found in the community concepts of Siemens and Shell. Similar to the Hypertext

Organization of Nonaka and Takeuchi (1997), the community organization has

been integrated with business operations and competence centers in a feedback

Visualization of virtual Knowledge Communities 93

circle. Starting with the business processes, people work in either projects or proc-

esses and acquire practical experiences in their team work. Here, they generate

knowledge. When these practical experts simultaneously participate in communi-

ties, they connect to related subject-matter experts and ask for special business

problems or related issues and thereby validate their insights, transfer their solu-

tions to others and conjointly produce an integrated and disseminated best practice

approach. Special members of the communities are working for corporate compe-

tence centers. They try to elicit these emerging approaches and use them to create

guidelines for new projects and processes. These guides and documents get for-

malized and are fed back into the business operations as rules and methods for im-

proving business results. The circle of Organizational Learning is complete (cf.

Figure 19).

Figure 19: A practical Instantiation of an Organizational Learning Cycle.

This example utilizes the theoretical concepts of situated and social learning (cf.

Lave and Wenger, 1991) as a prerequisite for an organizational learning approach.

It recognizes that learning and knowledge creation happen in relationships be-

tween interacting people. Communities of Practice are thus a catalyst for the ex-

change of experiences gained in process and project activities - they provide a

mechanism which supports organizational learning.

5.2 Practical Examples for CoPs in Organizations

In enterprises, Communities of Practice play an important role as a complement to

existing organizational structures. They are (a) forming an important part of an

organization’s strategy, (b) creating new lines of business through innovations, (c)

Competence Centers

Capturing,

Standardization

and Publishing of

Knowledge

Function oriented

Project & Process Teams

Teamwork,

Practical experiences,

Generating knowledge

Process oriented

Community

Diffusion and validation

of knowledge, based on

experiences

Network oriented

Knowledge Object

Repositories

Knowledge intensive

Business Processes

Knowledge intensive

People Networks

(Communities)

94 M.Trier

solving problems rapidly, (d) transferring best practices, (e) developing profes-

sional skills, and (f) helping companies recruit and retain talent. (cf. Allee, 2000).

This section will outline some practical examples to give an impression of how

communities are employed and embedded in a company.

As a first case study and extension of the organizational learning cycle introduced

in the previous section, Siemens’s approach to KM emphasizes the central role of

communities for practical KM. Centric to all other elements of their Knowledge

Management System (KMS), they chose Knowledge Communities as their pri-

mary focus. This is a very innovate and insightful decision, which led them to em-

ploy ‘Communities of Excellence’ (Enkel et al., 2002). There, virtual groups focus

on functions, like process-engineering teams in the production or software engi-

neers in the development division. The groups have members of the respective

topical areas, processes, and projects. An IT-platform is utilized, containing dis-

cussion boards, ‘urgent request’ facilities, member directories, chat features, a

search functionality, news pages, and link collections. Official coordinators have

been established and are responsible for tracking the flow of contributions to de-

velop their subject area. Next to this individual contribution of various practical

insights, members have bi-annual meetings and special community projects. In

this way, the Virtual Community is enriched and backed by personal contact.

Unlike Siemens, many other companies have (or would have) selected the knowl-

edge marketplace or in other terms Knowledge Management via IT-portals as the

central building block and primary project perspective. This often yielded in un-

successful post-project results of the KM system (also compare for chapter 2.2.2).

As a result, Siemens connects 1200 software engineers in the R&D department in

a topic-centric group. In their consulting branch called SBS (Siemens Business

Services) they integrate 2000 consultants. The most impressive community comes

hand in hand with the Sharenet application and includes 8000 sales people in more

than 50 countries. The right hand side of Figure 20 shows the ‘content’ or output

of these communities. It is divided into three groups. First, implicitly, CoPs con-

tain relationships, experts, communities and contexts. Then, experiences and im-

pressions are created and finally methods, structures and process models are for-

mally produced. Implicitly these three subsequent types of content demonstrate

the value chain of knowledge development in a community.

Visualization of virtual Knowledge Communities 95

Figure 20: Overview about Communities at Siemens. Source: Heinold (2000)

A further example for a major company that built Communities of Practice is

DaimlerChrysler’s Tech Club Concept. Here, product families like trucks, large

cars, etc. are linked to the engineering areas like chassis, electronics, etc. This al-

lows for topic oriented collaboration of engineers, working for different product

families at distant locations (cf. Figure 21).

Figure 21: The Architecture of the Tech Club Concept.

Similarly, Shell owns many small groups, which were mostly informal in origin,

with hardly any structure or facilitation (cf. Shell, 2001). In 1999, the small groups

were combined into global networks called Communities of Practice. These

groups are now quite large and have a formal position in the organizational struc-

Training 2500 in Belgium/Luxembourg

Siemens Learning Valley

Implicit

Relationships,

experts,

communities,

context, ...

Emerging

Impressions,

experiences, ...

(Lessons Learnt)

Formalized

Methods,

structures,

process models,..

(Best Practices)

Business process Knowledge Community Content

R & D 1200 software engineers,

industry automation groups

Consulting 2000 SAP, management and service

consultants

Manufacturing 1500 process technology engineers

Sales 8000 sales people in 50 countries

for customized solutions (ShareNet)

Procurement &

Logistics 50..200 community members of

Corporate Procurement and Logistics

Training 2500 in Belgium/Luxembourg

Siemens Learning Valley

Implicit

Relationships,

experts,

communities,

context, ...

Emerging

Impressions,

experiences, ...

(Lessons Learnt)

Formalized

Methods,

structures,

process models,..

(Best Practices)

Business process Knowledge Community Content

R & D 1200 software engineers,

industry automation groups

Consulting 2000 SAP, management and service

consultants

Manufacturing 1500 process technology engineers

Sales 8000 sales people in 50 countries

for customized solutions (ShareNet)

Procurement &

Logistics 50..200 community members of

Corporate Procurement and Logistics

Large Cars

Small Cars

Trucks

Minivans

Jeeps

Advanced Vehicle Engineering

Body

Chassis

Electrical / Electronics

Interior

Powertrain

Scientific Labs

Vehicle Development

Program Management

Energy Management

Thermal

Large Cars

Small Cars

Trucks

Minivans

Jeeps

Advanced Vehicle Engineering

Body

Chassis

Electrical / Electronics

Interior

Powertrain

Scientific Labs

Vehicle Development

Program Management

Energy Management

Thermal

96 M.Trier

ture. Example communities are sub-surface, surface and wells. Each CoP has 1500

to 2000 members. Additionally, smaller communities are dealing with issues of

knowledge sharing, of competitive intelligence and with Human Resource Man-

agement. The communities have so-called 'hub-coordinators' for facilitation. They

meet with other coordinators once in three to four months. They are responsible

for the coordination of all activities within the various communities. The raised

questions and discussions mainly deal with applying a colleague’s expertise to ex-

ceptional situations in the business processes, for example drilling methods. A

special department is analyzing all the semi-formal contributions and utilizes the

insights contained to produce new process standards for the whole enterprise.

Later, Shell augmented this communication oriented approach by a Learning Cen-

ter, which trains best practices to about 3,500 senior and technical staff every year

for an average 10 days of courses. Moreover, it set up Action Learning whose fo-

cus is on encouraging engineers to share ideas following a process of learning by

doing, through extensive use of videoconferencing in order to connect to recog-

nized experts throughout the organization.

In the IT sector, IBM is a further example of a sophisticated community approach.

They host about 20000 participants in 60 IBM communities (cf. Gongla and

Rizzuto, 2001). The groups are dealing with core competencies, like enterprise

systems management, application development, testing methods, and organization

change, industry sector competencies (automotive, chemistry, distribution, fi-

nance, health) or they are discussing market competencies, including e-business,

package integration, total systems management, mergers and acquisitions, and fi-

nally KM. IBM intends to assign a certain degree of independence to the groups.

This results in different characteristics in their range, scope, adoption of a concrete

taxonomy versus rather general categories, or structured review versus loose

workflows.

Among the internal tasks are managing the community’s intellectual capital, shar-

ing tacit knowledge, communications, socialization, membership management,

and content management. IBM further set up external processes, which include

incentive recognition, business strategy development and execution, and compe-

tency development. Finally, they developed measurement processes, which span

both, the internally and externally oriented community processes.

In their CoP approach, IBM established a small set of roles, including a commu-

nity leader, a core team with a router, who first reviews new submissions and as-

signs another member for their evaluation, and category owners, which are subject

matter experts in the different sub-areas who are responsible for topic and content

development.

Further organizations which consider Communities of Practice as a core element

of an organization’s KM approach are American Management Systems (AMS)

and the World Bank. They utilize these groups to generate ideas during electronic

discussions, which are then developed towards new products. Further, they use

Visualization of virtual Knowledge Communities 97

CoPs to effectively solve problems: using the expertise of communities, the World

Bank is able to quickly respond on requests from clients from various countries

including issues like how to carry out reforms in different sectors, or asking for

assistance in technical areas. The groups quickly identify best practices from dif-

ferent countries. Similarly, Buckman Labs has been able to reduce response time

drastically through effective knowledge sharing.

Many more corporate examples illustrate the successful application of this special

instrument of Knowledge Management, e.g. ChevronTexaco Corp., BP p.l.c., IBM

Corp., or Unilever p.l.c. (cf. Allee, 2000).

5.3 Benefits for Organizations

Although some benefits already have been implied by the introduction, definition,

and practical examples, a thorough review of the main corporate benefits will be

discussed in this chapter. Their recognition helps in deriving appropriate manage-

ment interventions to maximize benefits and control costs35. The benefits intro-

duced in this section include the short term benefits identified in the literature like

improved learning or transfer of tacit knowledge and the rather complex benefit of

supporting the establishment of Social Capital and trust.

5.3.1 Short term Benefits of Communities of Practice

There are various short term benefits, which the organization can achieve by em-

ploying a Community of Practice (cf. Wenger et al., 2002). It can be an arena for

problem solving and provides quick answers to questions. Hence, it allows for

quicker and cheaper information search because one can easily identify and con-

tact topic matter experts. By having discussed topics more informed decisions can

be made with a higher quality. This also enables people to approach new problems

more efficiently. The community is not just providing benefits to the organization

but also to its members. It helps them with their challenges and gives them easier

access to expertise. They can contribute their valuable insights to other topics and

can easier become visible and recognizable with their specific experience. The

connected expertise gives more confidence and backing in the tackling of business

problems. But also, it is motivating to talk to colleagues about topics of interest in

an institutionalized and legitimized way. It provides people with a sense of be-

longing to a group that together can work on innovations in a topic field. By this

exchange, people’s professional identities are being emphasized and reputation

can be built.

35 These will subsequently constitute a foundation for generating a measurement system in

chapter 7.2.

98 M.Trier

Members create a shared understanding of a topic domain, which is supporting the

transfer of knowledge. Knowledge Communities are more suitable to keep knowl-

edge in use (“alive”) as databases or manuals. The tacit elements of knowledge are

maintained and shared or are adjusted to local necessities. This makes communi-

ties a suitable means for introducing new members into an organization. They de-

velop competencies and carry new developments into the company. They are often

quicker and less inert as business units.

The impression of participating in the progress of a cutting edge development mo-

tivates members. Communities provide a ‘home’ and identity in turbulent times,

where business units, projects, teams and affiliations are continuously changing.

Further, in flat hierarchies Knowledge Communities constitute learning fields to

share and discuss new ideas between employees.

Lesser and Storck (2001) examined the link between Communities of Practice and

organizational performance. They identified four areas of organizational perform-

ance, which are determined by CoPs. These include a decreased learning curve for

new employees, a more rapid response to customer needs and inquiries, reduced

rework and ‘reinvention’, and the spawning of new ideas for products and ser-

vices. Further “communities play a significant role in the development of Social

Capital, which in turn influences organizational outcomes”. This beneficial prop-

erty of developing an abstract and qualitative capital or value for the company will

now be analyzed to improve the understanding of employing CoPs in an organiza-

tional context.

5.3.2 Social Capital

Significant attention is being paid to the development of ‘human capital’, that is

the education, skills and expertise of productive employees. “However, sociolo-

gists such as James Coleman, Ron Burt and Mark Granovetter argue that there is

much more to explaining the differences in individual success than individual

characteristics alone” (Lesser and Prusak, 1999). This is consistent with the find-

ings of chapter 3. What the authors identify is the complex corporate success fac-

tor Social Capital.

Here, Barney (1991) notes, that the Social Capital of an organization is a distinc-

tive collective property, mediated by individuals. The underlying idea is that in-

formation is not spread evenly across all actions. The access to resources or in-

formation depends on the social contacts (Coleman et al., 1966; Granovetter,

1985). The connections of a person do thus imply a value, as this person is poten-

tially capable to access the connected resources (e.g. other’s problem-solving ap-

proaches). This value is especially implied as individuals have to invest substantial

time and energy in developing these relations (Burt, 1992). Once established, peo-

ple also tend to keep with their current social relations in order to minimize trans-

action costs (Ben Porath, 1980). This cost perspective is also responsible for rela-

Visualization of virtual Knowledge Communities 99

tionship maintenance. If a new contact only yields a marginal increase in benefits,

it is likely to be abandoned over time to leave more resources and time for the re-

maining links (e.g. cf. Granovetter, 1974). In this context, Social Capital can sim-

ply be understood as the value derived from productive social relations, which ap-

pears in the form of improved information routing, resource exchange, emotional

support, and public goods production and mobilization. This implies a reorienta-

tion “away from immediate ability of people to complete tasks, toward the longer

term effects on their ability to act collectively” (cf. Resnick, 2002).

Social Capital complements financial and human capital (Barney, 1991) and can

be regarded as an integral part of an organization’s intangible assets (Pennings and

Lee, 1998). The relation-specific assets, knowledge sharing routines, and effective

relational governance mechanisms help firms to leverage their relational resources

for knowledge acquisition and exploitation. The network-based view on organiza-

tions thus extends the resource-based view and suggests that dyadic and network

relationships provide competitive advantage for a company (Lane and Lubatkin,

1998). They suffice the criteria of a corporate resource for sustained competitive

advantage, as it is valuable, rare, hard to imitate, and imperfectly substitutable

(Pennings and Lee, 1998).

To closer examine Social Capital and its relation to corporate performance, Tsai

and Ghoshal (1998) conducted an empirical study, which proved that Social Capi-

tal had significant effects on the levels of knowledge exchange and combination

within an organization and hence improved product innovation.

The underlying construct of Social Capital was proposed by Nahapiet and Goshal.

They define this construct as “the sum of the actual and potential resources em-

bedded within, available through, and derived from the network of relationships

possessed by an individual or social unit” (Nahapiet and Ghoshal, 1998:243).

The authors further identify social capital as having three interrelated dimensions:

structural, relational and cognitive (Nahapiet and Ghoshal, 1998:251).

Together these three elements foster the creation and sharing of Social Capital by

improving the access to parties for combining and sharing intellectual capital, the

anticipation of the value of this exchange, the motivation for exchange, and the

ability of the organization to adapt to the emerging needs.

The structural factor requires the formation and actual existence of informal net-

works between persons in order to enable them to identify people with required

resources.

The relational requirement is supporting the exchange between the individuals. It

addresses issues like trust, shared norms and values, obligation, expectation and

identification. Trust is the most complex element, and emerges in regular, honest,

and cooperative behavior. “Social Capital is the capability that arises from the

prevalence of trust in a society or in certain parts of it” (Fukuyama, 1995:26).

100 M.Trier

The cognitive aspect of Social Capital finally includes issues like common context

and vocabulary, which is supported by the use of common artifacts and stories.

Lesser and Prusak (1999) link this model of Social Capital to the concept of Com-

munities of Practice in order to show, how CoPs are beneficial for the develop-

ment of Social Capital of an organization. Looking at the structural dimension, the

employees set up and maintain informal personal relationship networks with oth-

ers having similar or complementary interests. Experienced colleagues can be ac-

cessed and included in the problem solving. The knowledge of employees can be

easier evaluated and people from the outside (like new employees) can be better

connected to the network.

CoPs foster interpersonal interactions and over time build a sense of trust and ob-

ligation critical for Social Capital development. They can test the trustworthiness

and commitment of others. The community develops its own ‘informal currency’,

common norms and values or conditions of payments (notion of obligation). As

people are organized around a common topic, they provide the required shared

vocabulary. Further, the CoP generates knowledge objects or artifacts and stories

which allow newer members to pick up social cues and develop commitment

(Lesser and Prusak, 1999).

Figure 22: A positive Feedback Loop between Social Capital and the underlying

and resulting Activities. Source: Resnick (2002)

Resnick (2002) identifies seven productive resources which are produced in a so-

cial network over time: communication paths, common knowledge, shared values,

collective identity, roles and norms, obligations, and trust (also cf. Figure 22). As

Lesser and Cothrel (2001) note, “these resources address the challenges of our

three critical social capital dimensions. The first, communications paths, is struc-

tural, the next four mainly cognitive, and obligations and trust clearly relational”.

Summarizing, Social Capital is a value created in a social network over time and

leads to the increase in partially measurable resources. This notion of resources

also directs the attention towards evaluating the actual extend of the utilization of

such values, or in other words: the measurement of the community. This idea will

be extended later in chapter 7, where among other dimensions Social Capital in

Social Capital

•Communication paths

•Common knowledge

•Shared values

•Collective identity

•Obligations

•Roles and norms

•Trust

Activities

•Info routing

•Resource exchange

•Emotional support

•Collective action

Social Capital

•Communication paths

•Common knowledge

•Shared values

•Collective identity

•Obligations

•Roles and norms

•Trust

Activities

•Info routing

•Resource exchange

•Emotional support

•Collective action

Visualization of virtual Knowledge Communities 101

Social Networks is being embedded in a comprehensive measurement framework,

which subsequently is being implemented in a concept for IT-based support for

evaluating Virtual Communities in chapter 8.

5.3.3 Trust

Related to the concept of Social Capital is the concept of trust and its benefits for

sharing knowledge in an organization: “Social capital is the capability that arises

from the prevalence of trust in a society or in certain parts of it.” (Fukuyama,

1995:26).

Fukuyama further defines the concept of trust as “the expectation that arrives

within a community of regular, honest and cooperative behavior, based on com-

monly shared norms on the part of other members of that community”.

In the literature, trust is identified as one of the most important and fundamental

success factors for people networks. It fuels sociability of the network as trustwor-

thy relations produce information benefits for the linked actors (Burt, 1992).

Here Powell states (Powell, 1996:226), that trust can come from shared socio-

cultural norms and traditions (‘characteristic-based trust’), institutions (‘institu-

tionally based trust’), and positive experiences with the cooperation (‘process-

based trust’; also cf. Zucker, 1986).

In the context of network oriented organizational structures (cf. chapter 3.2), Pow-

ell (1996) asserts, that networks require trust, as it enables them to handle complex

issues faster and more efficient as authority or negotiation. However, it has to be

noted, that the importance of mutual trust between employees is often only recog-

nizable, when common procedures have to be abandoned in times of change or

breakdown (Thomas et al., 2001). Then mutual trust supports collaborative effort

to proceed towards organizational as opposed to individualistic goals. Trust en-

ables people to build a more complex model of others as a basis for action in novel

situations. Thus, it helps to solve collective problems as it is reducing complex re-

alities much more efficient than projection, authority and negotiation and is imply-

ing that all participants restrain from free-rider behavior in the network and bring

in their share to conjointly generate benefits. Zucker (1986) and Lane and Bach-

mann (1996) identified three main factors, which support trust in a network:

• characteristic-based trust, which emerges from shared socio-cultural

norms and values,

• institutionally based trust, which emerges from a shared institutionalized

environment, and

• process-based trust, which results from positive cooperation experiences.

By fostering the development of higher levels of trust between employees by con-

tinuous contact and expressive informal communication, communities are eventu-

102 M.Trier

ally also providing a benefit to the whole organization, e.g. in the form of Social

Capital. Mutual trust is positively affecting the longelivety of organizations (De

Geus, 1997).

5.3.4 Understanding Costs to evaluate Benefits

If the benefits of a community are under concern, a thorough understanding of

their costs is a complementing prerequisite to arrive with a complete picture of the

contributions of Communities of Practice in an organization. Often, only the costs

of running Information Technology are taken into account as a cost estimate. This

leaves out various other cost factors like the costs of participation. According to a

recent empirical study by (Miller et al., 2002) on average, this factor contributes

52 percent of total costs. This share is mainly the sum of payments for members

with community related roles. Another 32 percent of total costs are caused by

face-to-face meetings. Usually this includes travel expenses and events but also

costs for online conferencing. Only 10 percent of total costs are actually attribut-

able to the provision and maintenance of IT. This can be explained with the fact,

that most companies do own the required hardware anyway and community soft-

ware is usually running on simple internet browsers (see chapter 6.2 on IT applica-

tions for CoPs). An average 6 percent are caused by the content management sys-

tem and promotion activities, like publishing contributions, maintenance of data-

bases or creating community newsletters (Miller et al., 2002:4).36

In summary, the previous sections showed that – offsetting occurring costs –

Communities of Practice provide a wide range of benefits for a networked organi-

zation and are thus an appropriate instrument for network oriented Knowledge

Management. The next chapter thus analyzes their elementary properties to iden-

tify, how their organizational structure works and how they can be set up and in-

fluenced in order to provide a value for the company.

5.4 Structural Properties of CoPs

A community’s organization can be analyzed in terms of its structural elements,

the established role model and the executed processes of the people (cf. Figure

23). These three domains of community structures will be described in the next

chapters. This section will start with the introduction of relevant structural ele-

ments.

Of course, identifying or defining a CoP structure requires a minimum amount of

formalism, which usually appears as a consequence of reaching a certain size of

36 The percentages where derived from a study of the IBM Institute for Knowledge-Based

Organizations, which analyzed 9 communities in 8 enterprises in the sectors financial

services, pharma, software, chemistry, telecommunication and tool manufacturing.

Visualization of virtual Knowledge Communities 103

membership. Building on the extensive work of Schoen (2000:94) about the em-

ployment, and organization of Communities of Practice in a corporation, these

structural constituents can be differentiated into rules and properties. Whereas

rules can be set and changed by the manager or coordinator of the community,

characteristic properties are emerging over time (more or less influenced by the

predefined rules).

Figure 23: Elements to characterize a Community Organization.

Structural properties are either emerging or are defined by the core group of the

community. They can be employed to influence the characteristics of the group.

First, the size and the members of the core group have to be defined. Schoen

(2000:95) suggests an optimal group size of five to six persons and proposes an

absolute maximum of 25 persons. Next to the core group, the overall group size

can be defined. Alternatively the group is left open to allow for unlimited mem-

bers. In the beginning, there can be a pre-selection of sub-groups or hierarchies

(Schoen, 2000:120). The membership duration, the continuity of their participa-

tion, their fluctuation and the actual growth of the group are related properties,

which can be measured. The overall duration of the group and the lifecycle stage

are the next indicator for CoP characteristics (also see section 5.6). This is fol-

lowed by the definition of the geographical spread and coverage of the target

group. Next, the type of the platform is determined together with its offered chan-

nels of communication (Schoen, 2000:96). Here, the mix of channels can be se-

lected and defined to fit the culture and context of the community. For this prop-

erty, there are various alternatives, which require individual processes: There are

channels which support one to one communication with a high degree of privacy.

Alternatively, there are the modes of one to many, many to many or few to few

communication (Schoen, 2000:123). The distribution of connected or affected or-

ganizational working processes and organizational hierarchies can be regarded as

a further constituting structural parameter. The community can be linked to exter-

nal units (like organizational divisions or other CoPs). The qualitative and quanti-

tative relationship structure itself is another defining property (Schoen, 2000:102).

For the perspective of this work, this last element is very important during the ac-

tive life of a community. The network-organization and the network oriented

COP Organization

Roles Structural elements

Processes

104 M.Trier

Knowledge Management introduced in the previous sections are both concepts

which imply that the actual structure of (network) relations is the fundamental and

dynamic property of a Knowledge Community. This outstanding importance will

be discussed in chapter 7 and is subsequently a major influence for the supporting

technical approaches introduced in chapter 8.

A final and somewhat more abstract underlying property is the actual rigidity of

these structures (or rather their dynamic behavior). This can also be measured,

evaluated and even be purposefully influenced by coordinators of the community

(Schoen, 2000:120).

The second category of structural properties consists of rules. These rules can be

actively defined and manipulated to change the properties above. Among them are

the rules of participation, the definition of the employed language(s), the intended

utilization of the outputs, the design of internal knowledge processes (e.g. ex-

changes, editing and documentation, review cycles), and the purposefully staffing

of certain community roles.

In a corporate setting, these rules can include the design of the incentive system

(i.e. the evaluation, who gets rewarded or disregarded for what actions), the setup

of the community strategy together with its system of business objectives and

monitoring. Finally it can be decided whether the community receives budgets and

in turn commits itself to pursuing external interests.

Figure 24 summarizes the structural elements of Communities of Practice.

Figure 24: Structural Elements of a Community of Practice.

Participation Rules

Language

Utilzation of Output

Knowledge Process Design

Role Allocation System

Incentive System

Target System –

Measurement

Budgeting System

Size

Number of Sub-Groups

Size of Sub-Groups

Membership Duration

Growth

Life Cycle Stage

Fluctuation

Target Group Coverage

Geographical Spread

Communication Channels

Organizational Spread

Hierarchical Spread

Participation Continuity

Relations to ext. Units

Properties of Relationships

Rules

Properties/

Characteristics

Structural

Elements

Visualization of virtual Knowledge Communities 105

5.5 Roles in a Community of Practice

Next to properties and rules, another important aspect of understanding CoPs is,

that the members of a community take diverse roles, which interact and conjointly

produce the network organizational structure called Community of Practice (com-

pare for example Schoen, 2000:117). The role model is usually constituted of a

core group of the most active members of the CoP. They often have a similar

function as a senior board, e.g. directing the community and making decisions.

Further, there are active members and usually a majority of mainly passive mem-

bers (Schoen, 2000:95). Here it should be noted, that despite their being inactive,

they are very valuable for the organization as they read contributions and learn

from the discussions of the topic matter experts. This is often not taken into ac-

count, when the active persons are measured against the passive members. The

members can also be differentiated into internal and external members. Here, the

latter are being affiliated to a separate division or organization (Schoen, 2000:95).

Further the duration of the membership or the experience can aid in separating

newcomers and experts. External to the community, there are potential members

to be acquired.

Figure 25: Overview about Community Roles.

Another role is the broker that connects similar people, who have not yet discov-

ered each other and their complementary background themselves. The manager

(also moderator or facilitator) is mostly one person in a coordinating position. The

actual name for this role should be carefully selected in practice in order to ex-

press the special intention of employing such a coordinator. The manager is the

most arguable, yet most powerful and authoritarian name for this job, whereas the

term facilitators emphasizes the aspect of servicing the group. Next to the man-

ager, there can be sponsors who want to achieve some business outcome of the

Members

(Active/Passive)

Active Members

Core Group

Broker

Manager

Facilitator

Moderator

Manager

Facilitator

Moderator

Coach

IT Support

Sponsor

Potential

Member

CoPOrganization

106 M.Trier

community work and therefore provide financial incentives in exchange for a us-

able output or hint in a specific topic. Last but not least, there are coaches coming

from central KM divisions and employees responsible for IT-Support as a service

role.

The role model itself only comprises the structure of the community. However,

much more relevant are the activities of the members. For example, people share

tips like how to prepare a project offer for a certain technological solution. They

discuss their approaches in discussion boards, work conjointly on generic docu-

ments (like for example a generic prognostic instrument for estimating a type of

project risk), they maintain personal profiles and websites or work on think tools

to tackle specific business problems or topic issues, like for example a KM audit.

Some of these activities can even be structured and designed in order to allow for

a smooth and unified execution. In that way, a CoP even shows some simple in-

ternal business processes. An overview about typical such activities of members is

given in Figure 26.

Figure 26: Typical Activities of Community Members.

For the network oriented KM approach, which is based on Communities of Prac-

tice, the manager role is very important for the progress of the group as it is re-

sponsible for all coordinative tasks. Hence, it strongly relates to the task of

Knowledge Management as the person is overlooking the work of the experts and

intervening into the activities of the members. Given this important position the

CoP manager has to be supported in a systematical way. However, in previous

chapters of this book, the tremendous difficulty of formalizing or even steering

networks of people has been discussed. The question is, whether management is a

constructive or a destructive activity in the domain of community work. This im-

portance and huge potential which resides in a concept for management together

with the specific difficulties to actually realize the potential benefits in practical

applications is a main reason, why this role needs to be analyzed in more detail.

Therefore it is now selected and discussed in order to identify concrete potential

for conceptual and technical support.

explain their work

discuss their needs

discuss their aspirations

share information

share hints & tips

share advices

share insights

help each other

consult each other

discuss their approaches

discuss their expectations

solve problems

explore common issues

create tools

create standards

create generic designs

create generic documents

organize documents

maintain a Web site

develop trust, understanding

develop (negotiate) common

perspectives, approaches,

practices

Visualization of virtual Knowledge Communities 107

5.6 Management of Community Structures - Pro and Cons

The examples of Siemens, DaimlerChrysler, or Shell illustrated professional ap-

plications of communities in enterprises. In such a commercial approach, often a

coordinating role is established as an organizing and steering contact person to ac-

count for the increased responsibility of the group.

A primary intention for the employment of such a management role in a Commu-

nity of Practice can be seen in the legitimization of the group in the organization.

Here, Wenger (1998) introduces a maturity model, which shows different degrees

of legitimization. In the beginning, the community is not recognizable for its

members (level unrecognized), subsequently it becomes an entity for a central

core of people (level bootlegged), in the third level the CoP is finally recognized

as an institution in the organization (level legitimized).

Degree of Formalization Description

Unrecognized Invisible for the organization and to some extend

also for the CoP itself

Bootlegged Only informally visible to a limited amount of peo-

ple in the immediate environment

Legitimized Officially accepted as a valuable unit

Strategic Widely recognized as having a central role for cor-

porate success

Transforming Able to influence and redefine its corporate envi-

ronment and the whole orientation of the organiza-

tion

Figure 27: Maturity Stages of a Community of Practice. Source: Wenger (1998)

To deal with the competition for time and resources with the project or line or-

ganization, or for improving the development of competences, the CoP Manager

can employ outbound oriented tasks like the active communication of the exis-

tence and the results of his community. This role hence needs to steer the group

towards a synergy (i.e. resonance) with the organization’s objectives. This implies

to select and pursue topics, which are of relevance to the enterprise. Further, the

management can employ sponsored resources and meet the controlling needs of

the respective sponsors (North et al., 2000).

The most important property with consequences for the feasibility of a manage-

ment role and the subsequent design of coordination measures is the degree of vol-

untarism and bottom-up orientation of CoP-members. Many community mecha-

nisms and platform activities can be traced back to internal group dynamics and

108 M.Trier

social mechanisms. Therefore, the flexible and self-organizing communities are

rather autonomous and difficult to manipulate and control by external persons

(Stamp, 1997). Some authors even strictly neglect external exertion of influence

(Por, 1997).

However, there is a scientific group which – despite accepting the difficulties –

shares the opinion that communities can be supported and hence also be controlled

by new means of work, despite the voluntary work of their members (Wenger,

1998; McDermott, 2000; Gongla Rizzuto, 2001). For example North states, that

the contexts for living Knowledge Communities can be purposefully designed and

created (North et al., 2000).

The proponents of this view highlight the personal leadership qualities, which are

necessary for successfully establishing working communities (Henschel, 2000). A

special aspect is the acknowledgement of the leader, which has to be rooted in his

work within the community, e.g. by his expertise or organizational engagement,

but not by any disciplinary definition (Wenger, 1998).

Figure 28 summarizes the main arguments for and against a management of com-

munities. On the one hand, the management has to define certain selection criteria

and rules for their members although communities are based on voluntarism. Fur-

ther, communities need shared and lively events and meetings to keep up the con-

tinuity and motivation as well as to strengthen the trust despite the advantages of

applying virtual interaction platforms. The CoP shall discuss topics relevant for

the company, but still the members need to establish a high degree of involvement

and personal interest. A community is being formed by certain situational needs

and should be able to adapt to changing environments (e.g. size, members). Never-

theless, it also requires a certain amount of continuity. Handling and utilizing

these major trade-offs is simultaneously the major challenge for management.

Visualization of virtual Knowledge Communities 109

Figure 28: Management of Communities: Pro’s and Con’s.

The following argument further describes this seemingly contradictory situation:

"Communities of Practice do not usually require heavy institutional infrastructures,

but their members do need time and space to collaborate. They do not require much

management, but they can use Leadership. They self-organize, but they flourish when

their learning fits with their organizational environment. The art is to help such com-

munities find resources and connections without overwhelming them with organiza-

tional meddling. This need for balance reflects the following paradox: No community

can fully design the learning of another; but conversely no community can fully de-

sign its own learning.” (Wenger, 1998a).

The special challenge of a moderating or managing role is also widely discussed

and characterized in the literature. Examples are McDermott’s community leader

(McDermott, 1999) or Fontaine’s concept of a facilitator (Fontaine, 2001). By

analyzing existing communities and their success factors on a more detailed level,

Kim develops a seven role model including the three roles host, event coordinator,

and greeter (Kim, 2000). A similar concept is found by Wenger (Wenger, 1998)

who even identifies seven leadership roles. Among them, there is an institutional

leader who is the link to the organization, an interpersonal leader who supports

tight social networks between group members, and a day-to-day leader, organizing

activities. In the end, all these roles can also be interpreted as special organizing

tasks, which can be attributed to a more general organizing role, referred to as

‘Community Manager’. This person takes the responsibility for the coordinative

tasks and represents the community at its external interface to the surrounding

formal organizational structure. While in some groups this role is very defined and

visible, in other groups the leader does not appear explicitly or its tasks and re-

sponsibilities are dispersed across various persons.

On the one hand... But on the other...

participation is based on voluntarism

and personal interest and

engagement of the individual member

the management has to define

certain selection criteria and rules for

their members

communities can communicate using

cheap virtual interaction platforms

like intranet or mailing lists

communities need shared and lively

events and meetings to keep up the

continuity and motivation and

strengthen trust

the community shall create a high

involvement and emotional binding of

the members

the CoP shall discuss topics relevant

for the company

the community has to provide for a

certain continuity to create the

necessary trust between members

and to form a common knowledge

base.

a community is being formed by

certain situational needs and should

be able to adapt to changing

environments (e.g. size, members)

On the one hand...On the one hand...On the one hand... But on the other...But on the other...But on the other...

participation is based on voluntarism

and personal interest and

engagement of the individual member

participation is based on voluntarism

and personal interest and

engagement of the individual member

the management has to define

certain selection criteria and rules for

their members

the management has to define

certain selection criteria and rules for

their members

communities can communicate using

cheap virtual interaction platforms

like intranet or mailing lists

communities can communicate using

cheap virtual interaction platforms

like intranet or mailing lists

communities need shared and lively

events and meetings to keep up the

continuity and motivation and

strengthen trust

communities need shared and lively

events and meetings to keep up the

continuity and motivation and

strengthen trust

the community shall create a high

involvement and emotional binding of

the members

the community shall create a high

involvement and emotional binding of

the members

the CoP shall discuss topics relevant

for the company

the CoP shall discuss topics relevant

for the company

the community has to provide for a

certain continuity to create the

necessary trust between members

and to form a common knowledge

base.

the community has to provide for a

certain continuity to create the

necessary trust between members

and to form a common knowledge

base.

a community is being formed by

certain situational needs and should

be able to adapt to changing

environments (e.g. size, members)

a community is being formed by

certain situational needs and should

be able to adapt to changing

environments (e.g. size, members)

110 M.Trier

Although the name of the role implies that communities are directly manageable,

this task is very special because of the principle of voluntarism in such networks.

There is no hierarchical power, which could be employed by a community man-

ager. Members dislike to be instructed and rather feel like a group of volunteers

who contribute their insights to a topic only, if they need to do it. This renders

management more a facilitating context management, which enables members to

work on their ideas (Fontaine, 2001). It is not focused at the distribution and com-

pletion of work tasks but on social interaction, learning, and knowledge exchange,

which subsequently and indirectly foster the development of solution and the work

on assignments. The generation of a strong identity and the emphasis on relation-

ship networks is next to content-related work a very important factor for managing

such a CoP. The manager has thus to indirectly and very carefully utilize the

mechanisms of a voluntary expert group. This has to be reflected in the tasks of

community management, which are discussed in the next section. Johnson (2001)

attributes this effect to the various constructivist properties of Communities of

Practice. They involve ill-structured problems, learning in a context of real-world-

problems, shared goals, and the use of cognitive tools to organize knowledge. Ill-

structured problems cannot be solved by any individual alone and hence the in-

structor is changing towards a facilitating coach for guiding the learning and help-

ing the team to develop. This moves the control away from the instructors to the

group and a network of people emerges.

According to this special situation, CoP managers are often emerging from the

group and are equipped with strong expert legitimation to strategically and tacti-

cally be able to influence the community development.

Finally, a theoretical argument for why the management of people networks is ac-

tually vital draws back to System Sciences and to its proponent Luhmann

(1984:46). He describes a group of elements as complex, if, due to immanent ca-

pacity restrictions in the ability to connect, not every element can be connected to

every other. This immanent restriction also implies that the system (e.g. a network

between people) can not fully access and observe the internal complexity of the

system’s elements (similar to information hiding in object oriented software archi-

tectures). In other words, because of the condition that these elements have to be

complex to allow for their utilization for complex system’s tasks, the elements (or

their interfaces) are so specific, that their ability to connect to every other element

is reduced. Complexity of elements hence creates the necessity for selection.

Every complex task requires careful selection of relations between a system’s

elements. Thus, it is neither possible nor efficient to connect all elements or peo-

ple in a network as the people are specialized and complex themselves. Rather, it

is necessary to have the system or acting units like a manager select and direct the

development and establishment of the network’s relationship structure.

Summarizing, although the management of Communities of Practice is unlike

conventional management, it can be utilized as a means of coordination. However,

Visualization of virtual Knowledge Communities 111

the management role has to consider completely new aspects, as it can not easily

apply hierarchical authority. To some extend, CoP managers are thus pioneers for

modern management approaches which can cope with increased autonomy of em-

ployees in working teams which utilize multiple and non-deterministic means of

communication. The next chapter clarifies these specific options and according

methods for managerial intervention into a network of experts.

5.7 What and How to manage

After having substantiated the benefits but also the key challenges of managing

Knowledge Communities, further questions arise. For example: What should the

manager actually consider and where are his concrete levers for intervention to be

found. The research on Communities of Practice is proposing a wide variety of

design and configuration concepts. One approach is to configure the rules and ob-

serve the changes in the properties of CoPs. This strategy has already been dis-

cussed in chapter 5.4. Another opportunity to identify domains of intervention is a

systematization of general success factors and their translation into a conceptual

management approach. As an example, the approach of North et al. (2000) will

now be briefly introduced to prepare for the next chapter, which is moving from

soft and generic domains to a more operational level of actual tasks for members,

moderators, and the other roles.

The approach of North et al. (2000) discusses four important domains of interven-

tion and configuration for a Community of Practice. The first element includes the

persons, who are members of the Knowledge Community. They contribute several

factors which are simultaneously domains for affecting a CoPs configuration.

Such people oriented factors are their motivation, their organizational affiliation,

their level of expertise, and their diversity in terms of knowledge. A further do-

main is the interaction of these persons. Here, factors like communication forms

and intensity, the general atmosphere and the identity of a community are allo-

cated. The third domain is the resulting knowledge transformation from tacit to

explicit knowledge. Finally, the domain of organizational integration can be influ-

enced by changing the degree of formalization, the community borders and the

time horizon of the community.

112 M.Trier

Figure 29: Four Dimensions for designing and influencing Knowledge Communities.

Source: North et al. (2000)

If these domains are analyzed, it becomes obvious again, that the management of

communities has special challenges. It is focusing on communication, motivation,

knowledge development, and even such abstract concepts as identity. Together

with the discussion about the special challenge of managing a heterarchic people

network without legitimized hierarchical authority and the section on organiza-

tional benefits, it can be concluded that there must be a substantial dominance of

social motives of a community moderator. In fact, the literature about the question

of how to successfully running communities suggests a multitude of leadership

tasks. They are summarized in the following incomplete yet comprehensive list

(cf. Lave and Wenger, 1991; Nahapiet and Goshal, 1998; Hildreth et al., 1999;

Wenger, 1998; Schoen, 2000; Wenger et al., 2002; Thomas et al., 2002; Hemphill

and Westie, 1950):

• foster and maintain participation with valuable feedback ,

• balance solution exchange and solution development (in the form of con-

tent generation and integration),

• create group identity,

• create a setting of mutual obligation,

• integrate isolated participants,

• observe role formation,

• improve inefficient parts of the network,

• monitor quality of interactions,

Persons

Motivation

Belonginess

Level of Expertise

Diversity of Knowledge

Interaction

Intensity

Form of Communication

Athmosphere

Identity

Knowledge

Transformation

Tacit – explicit

results

Organizational

Integration

Degree of Formalization

Size Limits

Time Horizon

Persons

Motivation

Belonginess

Level of Expertise

Diversity of Knowledge

Interaction

Intensity

Form of Communication

Athmosphere

Identity

Knowledge

Transformation

Tacit – explicit

results

Organizational

Integration

Degree of Formalization

Size Limits

Time Horizon

Visualization of virtual Knowledge Communities 113

• refine and share best practices (i.e. via contents and communication),

• understand existing informal CoP structure,

• increase informal learning activities,

• foster innovations,

• create familiarity between persons,

• communication of purpose, objectives, progress,

• balance group autonomy vs. openness,

• increase importance for participants,

• create relationship networks with tight connections and transparent visi-

bility of members within the network,

• analyze specialization and roles of individuals to form role architecture to

increase group stability, because over time group work is improving,

• create mutual trust,

• foster and communicate homogeneity and similarity in groups,

• weaken detrimental factors like concurrence and unsupportive personal

profiling,

• create and environment of obligation and mutual commitment,

• analyze interaction and interactivity,

• create group stability, and

• influence orientation and objectives (polarization/diversity).

These management issues are very consistent with the literature on the role of

community managers. Next to the actual creation, integration, refinement, ex-

change, and maintenance of contents (as knowledge assets), the creation of people

networks appears as an important objective, again highlighting the difference be-

tween codification and personalization (cf. Hansen et al., 1999). It can be noted,

that they do not include the usual means of conventional management: the delega-

tion of tasks and control of results. It is more a context management, where the

person in charge of developing the whole group has the complex challenge of so-

cially intervening into a big group. The conflict between acting as a manager

which needs to achieve objectives and acting as a catalyzing moderator, which has

to remain ‘invisible’, is characterizing the necessary care when using the term

management (cf. the ‘facilitator’ of Fontaine, 2001). The complex (and to a large

extent systemic) environment of a community moderator and his soft tasks and

challenges play an important role in this book’s approach for managing network

114 M.Trier

oriented Knowledge Communities and the subsequent software for supporting vir-

tual expert networks.

Finally, it has to be mentioned, that a definite list or characterization of leadership

roles in the context of community structures is still an issue under research. It can

not be developed by applying conventional management theories and approaches,

as community coordination fundamentally differs from other groups, like e.g. pro-

ject teams. However, the issues discussed in this chapter provide a first orientation

for the difficult question, where the moderators can be supported in their manage-

rial tasks. Especially the strong sociological emphasis in such network oriented

organizations affects the complex issues of measurement, analysis, and evaluation

of the current state of the community, which is a major concern for CoP facilita-

tors.

5.8 Tasks and Processes during the Community Lifecycle

The previous sections outlined the importance of managing groups of people when

managing CoPs. This also relates to the issue of dynamic network evolvement.

Here, the formation of a Community of Practice is essentially a bottom-up phe-

nomenon. It starts with various small cells of maybe two or three people that in-

teract to solve business issues. They do not know which relationships the other

person has in the back. So the network is only partially known and it is very de-

centralized. These cells slowly grow in size, more people are incorporated, and the

whole extent of the network gets more transparent. People learn to whom their

colleagues address their questions. By this process, the network gets identified

completely, yet it is still decentralized and communication from one point to the

most distant opposite point of the network is difficult, time consuming and

unlikely. Sometimes, the network gets centralized in a next step, using a platform.

Now everybody reaches everybody via a central (often IT-supported) platform.

Here, the character of the network fundamentally changes, a thing that every

community coordinator has to consider. Often the close and exclusive social ties

to some important contacts lose in value as other people now can also easily ac-

cess this person. So, obviously there are disadvantages and advantages to consider.

To analyze the coordinator’s influence on his community more systematically, a

lifecycle concept can be established. Similar to Wenger’s maturity model, a life-

cycle perspective emphasizes the increasing development of the informal network

and models general development stages. It renders it possible to derive the neces-

sary demand for a successful cultivation of the underlying social network. By this,

the general tasks of the previous section can be applied to influence the evolve-

ment and progress of an actual ‘object’.

Visualization of virtual Knowledge Communities 115

In the literature several such lifecycle models are proposed (e.g. Wenger, 1998;

McDermott, 2000; Gongla Rizzuto, 2001). Two of them will be very briefly re-

viewed before an integrated and simplified life cycle approach is presented37. It

helps to allocate the multiple tasks for members, managers and other roles on a

time scale of evolutionary steps. These tasks can again help to increase the under-

standing of what sequences of behaviors do happen in a community in order to

shed more light on the question, how a suitable support should be designed for the

participants of such expert groups. However, as the people integrating tasks are

eminent throughout the lifecycle, these stages are also focusing on content-related

tasks.

5.8.1 Wenger’s Community Lifecycle

Wenger (1998) proposes a lifecycle model which is progressing through five

stages: potential, coalescing, active, dispersed, and memorable. First, people face a

similar situation without a shared practice and they start to find each other. They

initiate discussions about shared topics of interests to utilize each others practical

experiences. A decisive criterion for the emergence of a community is the inten-

sity of the mutual interaction depending on the necessity of collaboration. This re-

sults in the establishment of informal relationships.

In the coalescing stage, the employees come together and start to recognize their

potential. They have several meetings or other joint interaction and communica-

tion. They use these instances to explore their similarities and to define shared ob-

jectives. This leads to an engagement in conjointly developing a practice. It is not

completely established; rather the community is negotiating it. Every member

takes the position, that the engagement should also benefit its own objectives.

After having established a shared practice, a period of increased working activity

starts. New knowledge is getting generated and exchanged between participants.

Artifacts like documents or stories emerge. New topics move from the periphery

to the center. Relationships develop between CoP participants. Novel topics may

eventually lead to the development of sub-communities or even completely dis-

tinct groups.

However, over time the intensity of engagement declines, still the community is

the center of knowledge-related activities. Some people meet from time to time to

discuss current problems or look for mutual support. However, the importance de-

creases for members, and persons which do not belong to the active core even

completely stop their participation. This makes the community less attractive for

members and there are no newcomers. This chain reaction increases the decline

until the group enters the memorable stage, where the community loses its central

role, but its meaning for people’s identities remains persistent. Important artifacts

37 McDermott’s model strongly resembles Wenger’s model.

116 M.Trier

are sometimes taken for a subsequent community, so that individual parts of the

CoP survive.

A similar lifecycle sequence has been identified by McDermott (2000). It consists

of the stages plan, start-up, grow, sustain, and close.

5.8.2 IBM Lifecycle

A further lifecycle concept, which is actually more a maturity stage model, has

been identified by Gongla and Rizzuto (2001) in their observation of IBM’s

Communities of Practice. In the ‘Potential’ stage, a community is forming around

a nucleus. This core is consisting of individuals with a common work interest.

However, the group has not completely discovered what their commonality is.

Connections are the fundamental issue of this stage. People must be able to locate

each other, communicate and form relationships.

In the following ‘Building’ stage, the community defines itself and formalizes its

operating principles. Here, the members declare its existence. The core group

starts to create structures and processes for the community’s operations. Context

creation and memory are fundamental functions. Members start a shared history

by conjointly creating things, which later become utilized. This also fuels the

common understanding of the notion of membership. The core starts to reach out

to others who should join.

During the ‘Engaged’ stage, the community executes and improves its processes.

It operates with a common purpose and grows in size and complexity. Access be-

tween community members is a key function to leverage explicit and tacit knowl-

edge. The community learns about itself as an ongoing entity, e.g. how it adjusts

and improves.

Phase Definition

Potential A CoP is forming.

Building The CoP defines itself and formalizes its operating principles.

Engaged The CoP executes and improves its processes.

Active The CoP understands and demonstrates benefits from KM and

the collective work of the community.

Adaptive The CoP and its supporting organization(s) are using knowledge

for competitive advantage.

Figure 30: Community Life Cycle Model of Gongla and Rizzuto.

Source: Gongla and Rizzuto (2001)

Visualization of virtual Knowledge Communities 117

In the subsequent ‘Active’ stage, the community reflects, understands, and dem-

onstrates benefits from knowledge management and the collective work of the

community. It solves business problems and exploits business opportunities,

which over time increases the CoPs role for pooling knowledge. Further, it ex-

tends its membership and builds relationships to external groups.

Finally “Adaptive”, the community and its supporting organization(s) are using

knowledge for competitive advantage. It reacts to external conditions, innovates

and generates new business objects, like solutions, offerings, methods, processes,

and even new groups or trends in its area of expertise. Figure 30 summarizes the

lifecycle model of Gongla and Rizzuto (2001).

5.8.3 Integrated Lifecycle

Figure 31 integrates the three introduced approaches into a simple life cycle model

with four phases. Before communities get implemented, it is advisable to identify

preparatory steps in the context of the preparation phase (compare for McDer-

motts Planning Stage). Afterwards, the actual identification and formation of the

community follows in the build-time-phase. Subsequently using the measures in-

troduced in the next section, in the iterative and feedback-like run-time-phase the

maturity and performance of the CoP can be increased (for example by the intro-

duction of IT platforms). The objective is to postpone the final completion phase

(Wenger, 1998).

Figure 31: Four Phase-Model of a Community Lifecycle. Source: Trier (2002)

This general concept is now utilized to allocate the tasks and processes of com-

munity coordinators to the four stages. This gives a good picture of what actually

happens inside an expert group, which in turn enables to better understand, how

such groups can be supported by methods or even IT-applications.

Potential

Potential Coalescing

Coalescing

Plan

Plan Start-up

Start-up

Potential

Potential Building

Building

Active

Grow

Grow Sustain

Sustain

Engaged

Engaged Active

Active

Adaptive

Dispersed

Memorable

Preparation Build-time Run-out

Run-time

1) Wenger 2) McDermott 3) IBM

Potential

Potential Coalescing

Coalescing

Plan

Plan Start-up

Start-up

Potential

Potential Building

Building

Active

Grow

Grow Sustain

Sustain

Engaged

Engaged Active

Active

Adaptive

Dispersed

Memorable

Preparation Build-time Run-out

Run-time

1) Wenger 2) McDermott 3) IBM

118 M.Trier

However, it is difficult to find a clear separation between members’ and manage-

rial tasks in a community. Often, more than one leader emerges (e.g. compare the

seven leadership types of Wenger, 1998), which acted as a regular member and

might return to that level.

Next to the separation into lifecycle stages, another distinction can be drawn to

systematize the tasks. The manager’s position at the border between the commu-

nity and the formal organizational structure, the tasks of the CoP-coordinator have

two directions: there are inbound oriented and outbound oriented tasks. Inward

oriented tasks are primarily directed towards the creation and communication of a

group’s identity. This can be developed by a bottom-up approach, by which the

roles and working tasks are integrated and formalized. These tasks can also in-

clude the decision of organizational questions and the support of other coordina-

tive roles, like for example, moderators, knowledge stewards or editors, brokers,

or event organizer etc. (Trier, 2002). Outbound oriented tasks are focused at the

environment of the community, i.e. communication the progress and value genera-

tion to the external stakeholders.

During the stages of the community lifecycle different issues arise. They are now

discussed in the next chapters

5.8.3.1 Preparation Stage

For a successful implementation of communities a preparation stage should be

considered. Here, the necessary prerequisites are examined, evaluated and created.

Barriers for successfully running communities are primarily in the domain of so-

cial relationships and the corporate culture. A cultural assessment and an as-is-

analysis of the current informal processes can avoid the problem of many KM ini-

tiatives that in the end the necessary acceptance of the solution may not be

achieved. This can happen, if it has not been invested enough upfront in order to

create a satisfactory understanding for the existing working processes and modes

of the employees (Allee, 2000). In the underlying social networks the fundamental

principle is trust (also see section 7.1.3 for more). Trust is especially important, if

in large Virtual Communities employees meet new colleagues to create exchange

relationships. Next to the management of organizational conditions, further pre-

paratory steps are advisable, like for example the identification of the individual

benefits of the community under consideration and its suitable topic areas or a

community development strategy (Wenger, 1998).

According to Wenger, this strategy determines the subsequent requirements for

the various organizational divisions. Line managers need to provide the necessary

time and resources to ensure, that the persons can participate in suitable and bene-

ficial communities in order to utilize the gained insights for their line projects. The

corporate strategists have to create a bi-directional channel between strategy for-

mation and community management. Finally the work environment must be de-

signed accordingly.

Visualization of virtual Knowledge Communities 119

5.8.3.2 Build-time Stage

During the build-time stage, a core of individuals with shared interests engages

repeatedly in informal and problem oriented interaction. They start to learn about

each other and get familiar (Gongla and Rizzuto, 2001). In this critical process, the

options for management intervention are restricted to the identification of the ex-

isting tightly connected core and its extension with further suitable and comple-

mentary persons. For this activity, especially the informal interaction between em-

ployees needs to be studied38. It can also be necessary to start with uncovering and

exploring the similarities of a group of related persons (Wenger, 1998).

For that, the usually inherent desire of emergent communities to grow (Nickols,

2000) should be supported with an appropriate organizational service, which can

e.g. organize meetings or supply necessary resources. Schoen (2000:143) suggests

conducting a preparatory kickoff-workshop. It is usually participated by a core

group. Such workshops clarify the demand for an output and the required compe-

tencies. Further, the group explores its identity as well as methods and instruments

for their community work. At this point, Wenger suggests the conception of an

enterprise-wide community awareness campaign (Wenger, 1998) which is an out-

bound directed task to inform the organization about the existence of the CoP and

about the likely values it will generate.

In sponsored CoPs (i.e. initiated by management), business benefits are derived

from the organization’s objectives and persons with the appropriate profile get

contacted. This scenario is strongly related to team structures with dynamic pat-

terns of interactivity but without a fixed set of members or fixed tasks.

In the next step, the community starts to form an own identity. The members gen-

erate defined processes and start to create structures. Typical roles are formed, like

the enthusiastic champion, the integrating facilitator, or the competent expert

(Nickols, 2000). The members create first artifacts and the history of the group

starts, which slowly forms the actual purpose of the group over time. With this,

the desire starts to contact and recruit new external members. The management

activities of this stage include the provision of help to find a group’s identity as

well as the facilitation of the decision to become a visible and recognized commu-

nity. Further, management can help with the planning of necessary processes and

structures, like roles or codes of conduct, help with the identification of new

members and finally it can deal with the capturing and documentation of knowl-

edge flows and the integration of problem solving strategies.

After the structure and the institution of membership have emerged, the designed

processes are executed in practice. The size of the community increases, it be-

comes self-sustaining and it learns about its own existence and characteristics.

38 This investigation can also be supported with the tool developed and introduced in the

final chapters of this book,

120 M.Trier

First learning processes appear and are implemented as improvements in the line

divisions. The members share their experience and support each other in their

business processes which are external to the community. The manager supports

the integration of generated artifacts into a structured knowledge base (Knowledge

Management System), the socialization of members, the establishment of feed-

back structures and processes as well as means of introducing, motivating, and

monitoring a result oriented community work (cf. Jarvenpaa and Leidner, 1998).

At this point in time, the migration to an IT-supported platform is especially for

emergent Communities of Practice a very fundamental intervention, which also

has consequences on the subsequent management of the group. Important decision

criteria are the culture of the participants and the expected number of members as

well as their expected geographical dispersion. This topic is introduced in more

detail in chapter 6.

IT-platforms first start to affect the necessary condition of (ideally personal) social

and not only professional connections between employees and therefore can cause

a high risk in certain community cultures. Despite this risk, it is usually very bene-

ficial to employ IT-platforms to guarantee the efficiency of the group with grow-

ing group size. Additionally, the community is given a (virtual) center, where al-

most all information can be seen and accessed. This is the first time, that all inter-

action can be observed explicitly, which improves the recognition of the individ-

ual identity. From a management perspective, this visualization and communica-

tion of the joint identity is a critical success factor. Coordinators should thus con-

duct trust-forming activities like the reduction of uncertainty about the employed

technology and processes, they should ensure expectable communication, e.g. by

notifying absences or introducing numbering systems.

5.8.3.3 Run-time Stage

Between the build-time and the run-time stage there is a continuous transition. The

content-related work is slowly replacing the formation of structure. People con-

duct main and related activities to generate the CoP’s output (Schoen, 2000:116;

also cf. Figure 32). Main activities are documentation, discussion processes, utili-

zation of the storage systems, cooperation on problem-solving processes or even

just-in-time support for other’s business problems, evaluation of contents, mediat-

ing and sharing contacts etc. External knowledge is acquired and the group looks

for potentials and benchmarks. Moreover the definition and management of han-

dled topics is being conducted (Schoen, 2000:140). Related activities of this stage

are maintaining contents, or identity-improving activities.

For every problem, dynamically a cluster of experienced and interested persons

emerges. They support each other and apply the shared knowledge in concrete

tasks outside the community. Reflexive feedback about the actual learning proc-

esses of those who applied the advices for solving their problem fuels the knowl-

edge and experience of the whole group. The manager can support this establish-

Visualization of virtual Knowledge Communities 121

ment of a feedback circle, which enables the adoption and improvement of the ex-

isting interaction patterns and working processes of the group according to the

new insights gained. Thus, with increasing maturity, the groups become adaptive

in their structure (Gongla and Rizzuto, 2001) and can start to autonomously incor-

porate changing requirements into their problem solving approaches. In this con-

text, a further important task of the facilitator is to bring together the capabilities

of the group’s members with relevant business problems of the enterprise. For

this, a main issue is to systematically nurture the communication. The latter is vi-

tal for knowledge processes and thus the management of communication channels,

behavior, and networks are a crucial part of the manager’s tasks. Knowledge Man-

agement hence boils down to communication management. This assumption plays

a major role in the final solution proposed in this book. Finally, the facilitating co-

ordinator has to communicate results and success stories like improved processes

or new solutions to the whole group and to external persons. This requires the

communication of value generation, the transfer of edited work results into public

documents and into knowledge bases, reporting processes directed to sponsors and

the alignment between the community’s and the corporate strategy.

Generally, the manager has to ensure, that the organizational interests are relevant

for the community’s work and has to avoid, that CoP members do engage too

much in work without significance for the enterprise (North et al., 2001). He has

to observe if barriers hinder knowledge exchange or participation and he has to

identify if the group is eventually approaching a decline stage.

An alternative approach for classifying the tasks of the run-time stage is a differ-

entiation into different knowledge related processes. This shows again the good

applicability of Communities of Practice for Knowledge Management. The com-

plete cycle of building blocks as proposed by Probst et al. (1997) ensures a com-

plete knowledge process. In community work, members document experiences,

identify demand for new knowledge, classify and sort new contents, edit and

evaluate the contents. Further they compress information and distribute it to suit-

able locations and persons. These knowledge processes are augmented by the

management tasks of the moderator, who is usually influencing the objectives and

observes activities and results.

These knowledge-related tasks of the manager differ according to the primary type

of the community’s interaction: This can be tending towards a personal face-to-

face group or comprising a primarily IT-supported virtual group. The first type is

usually found in a small group and the tasks of management are those of a team

leader with only indirect orientation towards the accomplishment of results. In the

usually larger Virtual Communities, the tasks of management are less bound to

personal contact and a rather content-related work area emerges39. This can gener-

39 Although the large group should additionally be treated as consisting of smaller groups.

122 M.Trier

ally be differentiated into three segments with a set of tasks for each (compare for

example Participate, 2001):

• Transformation of unstructured information, for example via tracking of

interactions to early discover and identify emerging information and con-

tent, establishing connections to qualification planning of the human re-

source division and to structured knowledge bases (Content Management

Systems, Document Management Systems, Knowledge Management

Systems).

• Diffusion of community knowledge, e.g. via newsletter or e-mail notifi-

cation, supply with relevant external content, ensuring ergonomic plat-

form functionality, adapt and improve the interaction and problem solv-

ing processes, edit contents for multiple reuse and wide application.

• Socialization (integration) of community members, using membership

programs, measurement of interaction and identification of established

social relationships (‘strong ties’) and key experts, connecting comple-

mentary persons and groups, creation of suitable incentives, organization

of events like off-site meetings of new members with CoP-champions or

membership management.

The measurable output created in the run-time of the community is primarily con-

tent-related or is represented in measurable improvements of the enterprise’s value

production. Examples for such innovations are:

• faster knowledge intensive process for setting up projects,

• more revenue per time period,

• widely available consulting expertise and best-practice approaches, or

• quicker achievement of full effectivity for new employees.

Internal Evaluations and measures can be utilized for an internal controlling and

an external reporting. The latter communicates the successes to sponsors, to CoP-

members, but also to potential knowledge carriers (experts) in the organization.

5.8.3.4 Decline Stage

In the final decline stage, a closing workshop to elicit and reflect and document

implicit insights and to prepare reusable output artifacts should be conducted. The

outputs are then handed over to the former members or to the enterprise. Some-

times the community transitions into a formal business unit.

The following Figure 32 summarizes the necessary tasks during the lifecycle.

Visualization of virtual Knowledge Communities 123

Figure 32: Overview about the main Activities in a CoP’s Lifecycle.

Figure 33 shows an overview about the introduced tasks of a community modera-

tor: In summary, the type of management processes is not only content oriented

but is primarily dealing with structural issues, which precede the content or

knowledge creation, e.g. the social structures and mechanisms of the group. The

role’s activities includes moderating processes (Schoen, 2000:140), initiation, ex-

ternal communication and connection to other groups, content maintenance, the

creation of transparency, the measurement and management of structural ele-

ments, topic management, and competence (and progress) assessment.

The table at the bottom of Figure 33 summarizes the predominant work domains.

Internally the moderator has to create value in three dimensions: People, Content,

and Network. He has to build expertise and qualification in the people domain,

identify useful contents and develop documents in the content domain, and foster

a working network of employee relationships. Then he has to communicate this

value to external stakeholders, including competence profiles, consulting topics,

documented knowledge, standards, stories or Best Practices, or communicate his

improvements in Social Capital.

Finally, from the multitude of activities shown in Figure 33, it can be implied, that

the management or coordination of a community is no easy venture. The manager

has to conduct various management tasks and has to influence the properties of the

community correctly. Further, he has to define the rules and to achieve external

and internal objectives.

Build-time

Run-time

Decline

Kickoff-Workshop

Implement Plattform

Designing Structure

Documentation

Discussion Process

Using Storage Systems

Problem Solving Cooperation

Evaluate Contents

Disseminate Contents

Maintain Content

Manage Structural Elements

Budgeting

Membership Management

Identity-forming Measures

Contact Mediation

Closing Workshop

Assemble Output

Handing Over to Users

Main Activities

Related Activities

Closing Activities

Starting Activities

124 M.Trier

Figure 33: Final overview about the tasks of a Community Manager.

In such a complex situation, conventional management is usually supported by

employing a generic management feedback loop consisting of a planning stage,

the subsequent execution of the plan, the measurement of the achieved results, and

the interpretation of the measures in regard to the last stage: adaptation of the plan

(e.g. cf. Hahn, 1996:41). This again emphasizes the important aspect of being able

to derive measures or find other means to evaluate the existing situation. Only

then, interpretation of this analysis can lead to improved interventions. The neces-

sity of such a measurement aspect is even more important, if the large size and the

limited transparency of expert groups with up to 1000 members are considered.

The aspect of measuring community structures, processes, and progress (evolu-

tion) as well as their impact on managerial tasks is the focus of the next chapter.

5.9 Role of Measurement for Managerial Community Tasks

The American Productivity and Quality Center APQC analyzed communities in

large companies. Here, systematic monitoring of effectivity and assessing the

‘health’ of the community has been identified as being a very important factor for

Knowledge Management in an enterprise. Next to the incorporation of the organi-

zation’s general strategic objectives and the leadership qualifications of the mod-

erating persons, the community structure is an important element of management.

This institution demands, that CoPs need to set up objectives and measure the ac-

tual performance using monitoring and controlling instruments (APQC, 2001).

For this, a manager has to be capable of systematically observe and evaluate his

group. As already introduced at the end of the previous chapter, ideally, he estab-

lishes a feedback loop of planning, assessment and evaluation. A central prerequi-

Executing Tasks

Initiation, Growth, Structural development,

Centralization, Developing Contents, Social

Community, Member Management,

Monitoring/Orientation, Coordination with Externals,

Commitments for budgets, Role Coordination

Executing Tasks

Initiation, Growth, Structural development,

Centralization, Developing Contents, Social

Community, Member Management,

Monitoring/Orientation, Coordination with Externals,

Commitments for budgets, Role Coordination

Achieve Targets

Improved Information Structure,

Info-quality as compared to the market,informed decisions, Expert

Transparency /-Contacts, Customer Value and Satisfaction, Train

new Colleagues, consistent procedures, bring knowledge to a

problem, standardized terminology, channel for idea generation and

validation, process improvements, disseminated Best-Practices, less

redundant work, additional revenue, faster Services, more reuse of

artifacts, Better learning curve, Amount of

knowledge/information objects, Reaction time

Achieve Targets

Improved Information Structure,

Info-quality as compared to the market,informed decisions, Expert

Transparency /-Contacts, Customer Value and Satisfaction, Train

new Colleagues, consistent procedures, bring knowledge to a

problem, standardized terminology, channel for idea generation and

validation, process improvements, disseminated Best-Practices, less

redundant work, additional revenue, faster Services, more reuse of

artifacts, Better learning curve, Amount of

knowledge/information objects, Reaction time

CoP

Manager

Influence Properties

Overall Size, Number of Sub-Groups, Size of Core Group,

Duration of Membership, Growth, Lifecycle Stage,

Fluctuation, Target group Coverage, geographical spread,

communication channels on platform, variety of

organizational processes touched, variety of hierarchy levels,

continuity of participation, relation to external units,

relationship properties

Influence Properties

Overall Size, Number of Sub-Groups, Size of Core Group,

Duration of Membership, Growth, Lifecycle Stage,

Fluctuation, Target group Coverage, geographical spread,

communication channels on platform, variety of

organizational processes touched, variety of hierarchy levels,

continuity of participation, relation to external units,

relationship properties

Design Rules

Participation, Language, Output Usage,

Knowledge Processes, Staffing Methods for

Roles, Incentive-Systems, Target and

Measurement System, Budgetsystem

Design Rules

Participation, Language, Output Usage,

Knowledge Processes, Staffing Methods for

Roles, Incentive-Systems, Target and

Measurement System, Budgetsystem

Content, secured and documented knowledge

> guidelines, Best Parctices, Stories

Content: Content Identification and

Development of Documents, Dissemination

Social Capital > measurable resultsNetwork: Social Relationship Network

Competence profiles, consulting, trainingPeople: Build expertise and qualification

Outward bound: Communicate valueInward-bound: Create value

Content, secured and documented knowledge

> guidelines, Best Parctices, Stories

Content: Content Identification and

Development of Documents, Dissemination

Social Capital > measurable resultsNetwork: Social Relationship Network

Competence profiles, consulting, trainingPeople: Build expertise and qualification

Outward bound: Communicate valueInward-bound: Create value

Visualization of virtual Knowledge Communities 125

site is the perceptibility of effects, which is usually n outcome of a measurement

system (as a ‘sensor system’ for a community). A further motivation to apply such

instruments is the need to report the community’s progress to the external organi-

zation, as described in section 5.6.

Next to APQC’s general demand for CoP measurement, other authors at least em-

phasize the need. A recognized thought leader in the field of Communities of

Practice, Richard McDermott, emphasizes the role of measurement for integrating

communities into the organization.

“For communities to become integrated into the business, they, like other organiza-

tional units, need to "play the game of business" and be judged against measures of

efficiency and effectiveness, like other organizational activities. So measuring value is

critical to fit into other organizational elements. The people who argue that Communi-

ties of Practice are too organic, too democratic, or too informal to be measured, do

more harm than good in trying to make Communities of Practice a recognized part of

organizations.” (Cachel and McDermott, 2001)

McDermott further notes, that all measures have stakeholders. A community

leader has different needs as a sponsoring business unit manager:

“A community leader, for example, needs to know what events and activities regu-

larly draw the most interest. So counting participation in meetings or hits to websites

can provide important information to them. But business unit managers typically have

little interest in this sort of information. They need to know if their investment of peo-

ple and resources are worth it. So they need measures of the value of the community.

Organizational leaders may also be interests in ROI information. But they may also

take a broader view, interested in how communities create or develop organization

wide capabilities or position them better in the market.” (Cachel and McDermott,

2001).

As these examples imply, the issue of measurement is a very important aspect of

community management. It helps responsibles to maximize the life-cycle duration

and the value generation of their expert groups. However, so far, there are almost

no profound approaches available. Although some authors propose preliminary

measurement systems (e.g. Cothrel, 2000), most of them only propose to count

clicks and documents in a software application. Usually, there is no systematic

consideration of the special social structures of such CoPs integrated and if it is,

then again no real measures are proposed. In this difficult situation, this book

wants to extend the current approaches in order to advance the discipline and to

improve the application of communities in enterprises. The special approach is to

analyze the communication of a group of experts. The underlying communication

contents and patterns are then employed to identify special social structures and

processes. These structural insights are translated into measures, which subse-

quently assist community moderators and managers in their design decisions. The

described approach takes into account the results of the discussion about network

organizations and the related requirements of Knowledge Management (compare

chapter 3.6). There, it has been identified, that network oriented KM primarily

126 M.Trier

needs to focus on (the support of) expert communication in a network of people.

Analyzing this communication to improve its support is thus a promising strategy.

With this approach in mind, looking for possible inroads for organizational but

also technical methods for community evaluation and analysis prerequisites a

closer examination of the role of IT and of community software in expert groups.

During the previous chapters, related issues like the special form of a Virtual

Community have already been touched. This aspect of IT-support together with its

implications for community management will now be discussed in more detail in

chapter 6. To anticipate the findings of the next chapters, IT plays a very impor-

tant role for the support of large groups of experts. This enables the establishment

of a technical approach to measuring communities in chapter 7, which is utilizing

the available electronic traces of communication for deriving suitable models of

the community infrastructure. The resulting approach is implemented as a soft-

ware (cf. chapter 8), which can help managers to increase the necessary transpar-

ency of their group. Some case studies illustrate this benefit in chapter 9. But first,

the important role of IT is compared to the actual offerings of community software

support.

Visualization of virtual Knowledge Communities 127

6 Information Technology to support Communities

The previous chapters introduced the environment and the main structures and

mechanisms of Communities of Practice. If such communities are to be supported

by means of Information Technology, all these aspects generate a multitude of dif-

ferent requirements for the software system. The following chapter analyzes the

role of Community Software, its development in the last year and in the future.

The main features are elicited and reviewed in order to give a picture of the state

of the art of current platforms from a Knowledge Management perspective. How-

ever, this evaluation will emphasize, that current software support ignores many of

the above requirements, i.e. the role of coordination and monitoring, the challenge

of identifiying and communicating benefits, or the aspect of life-cycle stages and

dynamic evolvement. Such identified gaps are the focus for the subsequent discus-

sion of a prototypical software application presented in the last chapter, which

aims at improving the IT support for Knowledge Workers and moderators.

6.1 The Role of IT for running Communities

Although communities do not necessarily need an IT platform for their work, it is

widely recognized nowadays, that IT can play a major role in efficiently support-

ing large groups of geographically dispersed experts (e.g. Hildreth et al., 1999).

This is especially important for international enterprises, where regularly similar

functions are spread across different divisions. Examples are local sales depart-

ments or decentralized product development departments using a component strat-

egy, like in the automotive sector. The most value of IT platforms is added by the

opportunity for one-to-many and many-to-many electronic communication over a

central virtual location. This implies fundamental challenges for a CoP Manager

when migrating from decentralized communities to centralized and transparent IT

platforms.

A good example for the increasing importance of IT support during the stages of

the community lifecycle is British Petrol p.l.c. In the beginning, they conducted

formal meetings in order to exchange expert knowledge. Next to such planned

events, a large number of informal and unidentified networks existed without any

rules. After the implementation of the community initiative, these groups became

visible and officially recognized (cf. the maturity stages model of Wenger, 1998;

see chapter 5.6). The identification of these groups increased public attention and

hence the relevant groups attracted more members and grew in size. Over time,

the members existed in geographically very widespread locations and face-to-face

contact became increasingly expensive. To compensate for the size, the communi-

ties were supported with a very sophisticated IT platform, which provided features

128 M.Trier

like mail centers, public folders, discussion boards, an integrated document stor-

age facility, and yellow pages (McLure-Wasko and Faraj, 2000).

The necessity of a central place for communication has also been substantiated

theoretically by Nonaka and Teece (2001). They established the concept of Ba,

stating that knowledge transfer always requires a place like in this setting the plat-

form in order to work. “Ba” is the Japanese word for place and represents the con-

text in which knowledge is created, shared, systemized, and exercised.

In order to utilize all these advantages of software infrastructures, the manager

needs to successfully migrate the very informal and invisible initial relationships

of his group of experts to this platform. However, the adoption and movement to a

platform has to be in line with the life cycle stage of the community (cf. Trier,

2002). The expert group originally emerges from informal relationships between

people, who start to develop a network without the application of information

technology. Over time, the growth in group size and the geographical distribution

of members directs the attention to the issue of technical support for these groups

and the application of a central community software platform together with related

service processes.

The main difficulty in employing software support is the change in network struc-

ture. A formerly decentralized network with many social elements is becoming

centralized on a platform. Persons with very exclusive relationships (sometimes

established over years) could be afraid of losing their special network position.

Moreover, the social character of the relationships is likely to be reduced, because

IT can only support social interactions between the members of a community, but

technology can rarely completely replace personal contact (Stamp, 1997) and its

important contexts necessary for establishing strong social relationships.

These adverse effects have to be compensated by the manager by means like face-

to-face meetings or the establishment of a strong and visible group identity.

CoP platforms are especially helpful for areas, in which tacit knowledge of ex-

perts can directly be applied to a related business problem (Brown and Duguid,

1998; Wenger, 1998; Wenger and Snyder, 2000). The people requesting help do

not need to tediously analyze documents and protocols of similar scenarios to find

and interpret a case with an appropriate fit to their problem. Instead, they can di-

rectly enter their request into a platform. A suitable subject matter expert can then

apply his existing knowledge to this special context and does not need to explicate

his experience into a broad and generic problem solution first. By answering ques-

tions of others and receiving the appropriate feedback about the practical imple-

mentation of their advice, experts are also frequently updated and reassess or even

extend their experience in new concrete application scenarios. For the initiator of

the request, this method is a better way to learn by applying other’s experiences.

Next to this ad-hoc mode of problem solving, community software provides the

community with a means to discuss, develop, and integrate distributed partial ap-

Visualization of virtual Knowledge Communities 129

proaches from projects or business processes to best practice standards. Communi-

ties of Practice are living longer than projects, which last only for a limited period

of time. This long-term perspective of topic oriented people networks helps the

organization to maintain important competencies achieved in various related pro-

jects even after they have been completed (Wenger, 1998). Experts generate their

insights in projects and can nurture and develop their knowledge in communities.

They can recognizably establish themselves as subject matter experts in a relevant

topic field. Additionally, a valuable archive of the members’ contributions is being

created.

6.2 IT Applications for Communities

On a technical level, communities in an enterprise mainly develop by following

one out of three migration paths (Trier, 2003; cf. Figure 34). In the first scenario,

the community platform develops from the initial application of groupware to

support teams in various corporate projects. These tools are becoming modified to

host defined topics and support the new user group of CoP-members. Afterwards

they are offered to emerging CoPs as an internal service. In the second scenario,

the organization decides to officially align the existing expert networks and targets

at connecting relevant employees without introducing a central document-centered

system. When the company follows this strategy, it either develops typical CoP-

functionality for internal communication and networking or it implements targeted

software from a platform vendor.

Figure 34: Three Migration Paths towards CoP development. Source: Trier (2003)

e.g. Shell, Mummerte.g. PWC e.g. Volkswagen,

Microsoft

Classical document-

centered KMS Employee/

HRM System / KMS Groupware

Migration

from the classical

(document -

based) KM-System

towards

identifying KMS-

User-Communities

which utilize the

platform as

medium

plus

extension of the

platform

Pure people-

centred KM

System

with ad-hoc

character, without

document

knowledge base

with improved

integration into

the organization

and improved

Controlling

Originated from

Groupware

and reinterpreted

as a COP-platform

with support of

the whole

management-

system

with improvement

of the integration

in the organization

and improved

Controlling

CoP

CoP-

-Platform +

Platform +

CoP

CoP-

-Management

Management

130 M.Trier

In the third scenario, the enterprise already adopted the codification strategy (Han-

sen et al., 1999) and runs a conventional primarily document-based knowledge

management system (KMS). This system is being utilized by various informal

groups of users. Although initially, the grouping of users is not directly intended,

they form invisible communities because of their identical interests and the estab-

lishment of various relationships over time. Often companies broaden their ap-

proach towards the personification strategy (Hansen et al., 1999) to directly con-

nect their employees and reduce the problems arising from maintaining large vol-

umes of documents, often referred to as knowledge objects. To identify and ac-

tively support the existing groups, corporate KMS’s are becoming enriched by

special community features for direct communication between the experts.

These multiple paths leading to IT support for expert groups already imply the

heterogeneity and dynamic development of this software segment. From various

related fields of applications, vendors are extending their product towards im-

proved community support. Examples for such moving market segments are

document-based knowledge bases and knowledge exchange systems, project

spaces and groupware, conventional discussion boards, tools for synchronous in-

teraction and internet-community software.

The different software segments and their development towards the support of

Communities of Practice are summarized in the following figure.

Figure 35: Software Segments integrate towards CoP Platforms. Source: Trier (2003)

This development towards an integrated product segment increases the risk of put-

ting too much functionality into one platform. This can result in detrimental com-

plexity effects affecting usage, e.g. training processes are taking longer, or re-

searching information takes more effort. Moreover, information exchange can get

inefficiently distributed over various communication channels (i.e. e-mail, discus-

COP-PlattformCOP-Plattform

Dokumentenbasiertes WMS

Knowledge Bases

Yellow Pages /

Expertise Sharing Software

E-Learning Software

Synchronous Interaction Tools

Electronic Discussiongroup Tools

Website Community Tools

Groupware / CSCW /

Project Spaces

Wissensportale/

Knowledge Worker Desktop

Extended by ad-hoc

Group Interaction

Extended by Group Interaction

and dynamic role-profiles

Extended by

Group Interaction

Extended by asynchronous

Interaction and Self-organisation

Extended by Document storage

And Self-organisation

Extended by Document storage

And Self-organisation

Extended by

Self-organisation

Extended by

Group Interaction

Dokumentenbasiertes WMS

Knowledge Bases

Yellow Pages /

Expertise Sharing Software

E-Learning Software

Synchronous Interaction Tools

Electronic Discussiongroup Tools

Website Community Tools

Groupware / CSCW /

Project Spaces

Wissensportale/

Knowledge Worker Desktop

Document-based KMS

Knowledge Bases

Yellow Pages

E-Learning Software

Tools for synchronous Interaction

Electronic Discussionboard

Website Community Tools

Groupware / CSCW /

Electronic Project Spaces

Information and Knowledge Portals/

Knowledge Worker Desktops

Visualization of virtual Knowledge Communities 131

sion group, instant messaging, and telephone). This segregates the expert groups

into sub-groups working on the same topic but missing each other because of a

preference for different communication channels. Hence, it should be considered

to prefer a simple solution with only some essential features and support for few

electronic communication channels according to the intended strategy for support-

ing the employees. This strategy should be rooted in the communication culture of

the organization.

The software to support Communities of Practice is no defined field, rather the

various features can also stem from different applications. But often enough, the

features are integrated into a special platform to be accessed by community mem-

bers. Features include discussion groups, calendar features, socializing gimmicks,

notification systems, blackboards, buddy lists etc. They do very much resemble

groupware applications and that is no surprise, as communities are similar groups

of people, working on a special problem. So it can be concluded, that not the pro-

vision of a unified tool is the innovation but the vitalization of the group in the

intranet.

The architecture for a community application is on an abstract level not very com-

plicated. It consists of a data layer, an application layer, and a communication

layer (cf. Figure 36). The application layer hosts all the various modules, like dis-

cussion boards or publishing tools. It can link to external modules and data

sources by using API’s (Application Programming Interfaces). The importance for

supporting the communication of the members is emphasized by the special com-

munication layer. There are many channels available for communication. Unless

all these channels are integrated, they have the tendency to divide communication

instead of making it ubiquitous. Examples for channels are e-mail, sms, facsimile,

pdf, browsers, wap-browsers or whatever their situation and location allows.

132 M.Trier

Figure 36: Generic architectural Model for a CoP Platform.

A sample architecture of a very renowned community software provider (Cassio-

peia) is shown in the following Figure 37. It contains all the elements of the ge-

neric architecture. It illustrates the connection to databases using the Java database

connectivity interface. They connect to user profiles using LDAP (light weight

directory access protocol). Further, the various backoffice-modules are shown, in-

cluding a data access manager, a user manager, a workspace manager and so on.

The front-office modules discussion, mailcenter, etc. are accessing these adminis-

trative programs.

Cassiopeia offers features like a discussion board, conferencing with synchronous

communication, expert requests, the creation of personal homepages without

knowledge about HTML by using master forms, publishing personalized pages,

data collections about the profiles of users, a mailcenter to send and receive mes-

sages, smart workflows for managing simple sequences of tasks, and toplists,

which rank contents using an evaluation and point-System (e.g. for contributions

to discussion boards).

SQL

Database

External

Data

Sources

User

Management

Chat Rooms

Blackboard

Discussion

Board

Mailcenter

Localizer

Publisher

Tracking

Homepage

Online-

Conferencing

API

External

Module

(php,c++,

Java,perl)

Templates (XML/XSL)

Web Server Gateway

User (Browser)

http/(d)html,flash

Administrator

(Browser)http/html

Mobile User

(Browser)wap/wml

E-Mail

SMS

FAX

PDF,rtf

Data Layer Application Layer Communication Layer

Visualization of virtual Knowledge Communities 133

Figure 37: Sample Architecture of Cassiopeia.

Another solution for community support, which will be briefly introduced, is of-

fered by Livelink. Although it also includes discussions, it rather concentrates on

attention management, which is the management of different views on relevant

information sources, project management with online project rooms, project status

reports, milestones, team management (e.g. add members and access control), task

management (task segmentation, task allocation, planning and monitoring), a data

base for project documents, and version control for shared documents. Further

there are embedded search mechanisms like fulltext- und index-based search to

provide access to complex organizational information- and knowledge object

bases. Only recently, Livelink has added a target set of features in its community

component for Livelink.

The applications Cassiopeia and LiveLink are being compared in the following

Figure 38 (Cassiopeia is on the left hand side, LiveLink is right hand side). The

grey bars represent functions employed by both applications. The boxes show spe-

cial areas, where the different approach of both applications become recognizable.

Light text represents unavailable features. This comparison shows that there are

only few standard features, which are provided by both applications. Hence, al-

though sold for the support of Communities of Practice, the tools are very differ-

ent. This comparison substantiates, that there is no standard tool for supporting

CoPs available, however some standard features slowly emerge (like e.g. discus-

sions).

Data Layer

3rd party

SQL

Database

3rd party

LDAP

Server

Application Layer

Administration

Manager

Workspace

Manager

InfoUnit

Manager

WebSpace

Manager

File Access

Manager

Function

Manager

Session

Manager

Data Access

Manager

Rating

Manager

Container

Manager

Event

Dispatcher

User

Manager

Request

Dispatcher

Buddylist

Calendar

Conferencing

Document

Discussion

Mailcenter

…

3rd party

module

CAPPHTTP

XML over HTTP

(XML Gateway)

External

Application

SQL LDAPData Broker XML Parser

Application Component API

JDBC LDAP

Templates

Style-

Sheets

Templates

Style-

Sheets

134 M.Trier

Figure 38: Comparing the features of Cassiopeia (left) and Livelink (right) as of 2003.

A third example is Communispace. Its features resemble groupware functionality,

including traditional features like discussion, chat, calendar, organizing docu-

ments, or personal profiles. But additionally, Communispace provides facilities for

specific content related activities such as framing issues, brainstorming, making

decisions, or analyzing the “community climate”. A brainstorming facility helps to

take the group through the various phases of brainstorming: generating ideas, dis-

cussing them, ranking them, etc. Communispace supports the reflection on the

quality of the community in terms of relationships, levels of trust and participa-

tion, nature of conversations, etc. Its discussion facility requests contributors to

categorize their contribution according to a taxonomy of ten different “speech

acts” including question, answer, request, offer, assent, dissent, etc. On the other

hand, the ability to handle documents is still underdeveloped and the search uses

keyword indexes only.

The open source system ‘ArsDigita Community Systems’ (ACS) includes five

domains: Personalization, Site Management, Collaboration, Publishing, and

Transaction. Every set contains a series of modules, which realize specific func-

tions. For example, the collaboration feature provides bulletin boards, discussions,

web-based intragroup e-mail, calendars, chat rooms or address books. Together,

Management Features

Archiving user data

Tracking

Manage members

Evaluation / Toplist

Personal Features

Expert (Request)

Database/Document

Search Functionality

Homepage

Publisher

Contacts

Calendar

Alert

News

Smartflows

Flexible user interfaces

„Attention Management“-Views

Group Features

ProfileAgent (YPage)

Group Awareness

Virtual Workplaces

Project status/project management

Discussions

Conferencing

Instant Messaging

Blackboard

Calendar

Buddylist

Presence Awareness

Mailcenter

Memberguestbook

Smartflows

Team management

Version management

Management Features

Archiving user data

Tracking

Manage members

Evaluation / Toplist

Personal Features

Expert (Request)

Database/Document

Search Functionality

Homepage

Publisher

Contacts

Calendar

Alert

News

Smartflows

Flexible user interfaces

„Attention Management“-Views

Group Features

ProfileAgent (YPage)

Group Awareness

Virtual Workplaces

Project status/project management

Discussions

Conferencing

Instant Messaging

Blackboard

Calendar

Buddylist

Presence Awareness

Mailcenter

Memberguestbook

Smartflows

Team management

Version management

Application BApplication A

Management Features

Archiving user data

Tracking

Manage members

Evaluation / Toplist

Personal Features

Expert (Request)

Database/Document

Search Functionality

Homepage

Publisher

Contacts

Calendar

Alert

News

Smartflows

Flexible user interfaces

„Attention Management“-Views

Group Features

ProfileAgent (YPage)

Group Awareness

Virtual Workplaces

Project status/project management

Discussions

Conferencing

Instant Messaging

Blackboard

Calendar

Buddylist

Presence Awareness

Mailcenter

Memberguestbook

Smartflows

Team management

Version management

Management Features

Archiving user data

Tracking

Manage members

Evaluation / Toplist

Personal Features

Expert (Request)

Database/Document

Search Functionality

Homepage

Publisher

Contacts

Calendar

Alert

News

Smartflows

Flexible user interfaces

„Attention Management“-Views

Group Features

ProfileAgent (YPage)

Group Awareness

Virtual Workplaces

Project status/project management

Discussions

Conferencing

Instant Messaging

Blackboard

Calendar

Buddylist

Presence Awareness

Mailcenter

Memberguestbook

Smartflows

Team management

Version management

Management Features

Archiving user data

Tracking

Manage members

Evaluation / Toplist

Personal Features

Expert (Request)

Database/Document

Search Functionality

Homepage

Publisher

Contacts

Calendar

Alert

News

Smartflows

Flexible user interfaces

„Attention Management“-Views

Group Features

ProfileAgent (YPage)

Group Awareness

Virtual Workplaces

Project status/project management

Discussions

Conferencing

Instant Messaging

Blackboard

Calendar

Buddylist

Presence Awareness

Mailcenter

Memberguestbook

Smartflows

Team management

Version management

Management Features

Archiving user data

Tracking

Manage members

Evaluation / Toplist

Personal Features

Expert (Request)

Database/Document

Search Functionality

Homepage

Publisher

Contacts

Calendar

Alert

News

Smartflows

Flexible user interfaces

„Attention Management“-Views

Group Features

ProfileAgent (YPage)

Group Awareness

Virtual Workplaces

Project status/project management

Discussions

Conferencing

Instant Messaging

Blackboard

Calendar

Buddylist

Presence Awareness

Mailcenter

Memberguestbook

Smartflows

Team management

Version management

Application BApplication A

Visualization of virtual Knowledge Communities 135

the five modules represent their model or framework of an ‘Online Community’.

The framework is shown in Figure 39.

Figure 39: Framework of the ArsDigita Community Application.

After having had a look on vendor products, the following Figure 40 shows a

screenshot of a developed system used by Infineon (a semiconductor manufac-

turer). It is obviously based on the Lotus Notes platform and contains features like

people, teams, events, activities, information, and discussions. In the main win-

dow, we can see a terminology taxonomy which is to support the search for topics,

discussion threads etc. It includes a mindmap of the whole division under consid-

eration, including process modules, document types, tasks, processes, technologies

etc.

All previous examples show, that different vendors and individual corporate de-

velopments claim to support communities but pursue a very different approach.

However, on an abstract level, community software applications have the follow-

ing core set of elements:

• CoP-Portal,

• communication-interaction-collaboration asynchronous: Discussion

Boards,

• communication-interaction-collaboration synchronous: work rooms,

chats,

• Yellow Pages: participants- and experts directories,

• CoP-Management-Tools: membership- and document management,

monitoring elements,

Structuring via an authoring system, Editing, Templates, Content Filtering, Publishing Systems,

etc.

Publishing

Membership Registration, Activity Tracking, Personalized Contents and Navigation, User

Profiles, personal Portals and Sub-groups, access control, etc.

Personalization

Access to information of various web-browsers, Bulletin Boards/ Discussions, Chat Rooms, Web-

based E-Mails, Calendar, Bookmarks, Adressbooks, File Storage, Presentations.

Collaboration

E-commerce-functions: „Recommendation Tracking“, Classification, Auctions, Security,

„Auditing“ and online Reporting.

Transaction

Directory, Statistics, Search, Protocols, User Requests

Site

Management

ModulesToolsatz

Structuring via an authoring system, Editing, Templates, Content Filtering, Publishing Systems,

etc.

Publishing

Membership Registration, Activity Tracking, Personalized Contents and Navigation, User

Profiles, personal Portals and Sub-groups, access control, etc.

Personalization

Access to information of various web-browsers, Bulletin Boards/ Discussions, Chat Rooms, Web-

based E-Mails, Calendar, Bookmarks, Adressbooks, File Storage, Presentations.

Collaboration

E-commerce-functions: „Recommendation Tracking“, Classification, Auctions, Security,

„Auditing“ and online Reporting.

Transaction

Directory, Statistics, Search, Protocols, User Requests

Site

Management

ModulesToolsatz

Framework

Site ManagementPersonalization Collaboration

Publishing Transaction

136 M.Trier

• forum for internals and externals to ask the community,

• document repository,

• search engine: find explicated knowledge in the documents and com-

ments or tacit knowledge by identifying relevant members and experts,

and

• a wide series of optional further tools: e.g. to create sub-communities.

Figure 40: User Interface of a corporate Community application.

To summarize the analysis of community software, it can be concluded that most

current features of community software focus on supporting search and retrieval

of contents and the communication between members. Next to such member ori-

ented features, the user group under research in this book, namely the coordinating

roles, like moderators, brokers or facilitators are still insufficiently recognized.

They do not only employ content oriented features, but need to observe, under-

stand, and communicate the development of their social CoP networks by means

of monitoring and analysis. Although some vendors already offer simple logging

facilities, current software is not recognizing the network oriented mode of

Knowledge Work and the impetus of social network theories for modern monitor-

ing software features. Hence an important potential can be assumed by extending

the existing functionality in this direction, as it enables the communication of CoP

development to different stakeholders. On an abstract level, Figure 41 shows the

main features of management oriented software facilities. They include the moni-

Visualization of virtual Knowledge Communities 137

toring of (social) group structures and activities, topic management, and report

generation.

Next to the support of community moderators these extensions would also benefit

the members, which get a better impression of their otherwise invisible group.

This fuels the community’s identity. A tertiary interest group is comprised of net-

work researchers and analysts, who frequently apply similar features provided by

statistical expert software to conduct studies of Virtual Communities.

Figure 41: Extending current Community Software with management oriented

Functionality. Source: Trier (2003)

6.3 Computer-mediated Communication in Virtual

Communities

The last chapter illustrated, that the employment of IT actually leads to an abun-

dance of new means of communication. However, this abundance also implies a

dispersion of meaningful dialog, learning, and knowledge across these channels.

Thus, the discussion of IT applications for the support of communities has to con-

sider the aspect that most Communities of Practice start to employ more than one

communication channel and that IT is not able to support all of them in one

transparent and integrated approach. Examples for such channels, which are not

integrated, are face-to-face communication or phone calls. So, every attempt to

systematically conduct network oriented Knowledge Management with communi-

ties involves this heterogeneous aspect of communication and the potential loss

of parts of the ongoing knowledge exchange.

Member-oriented Facilities

Management-oriented Facilities

PotentialFunctionality

Monitoring of (social) Group Structure

Monitoring of (social) Group Activities

Topic management (Portfolio)

Report-generation

...

Protocol-Data

Extended Functionality

Search Feature

Document Storage

Evaluation Systems

Buddylists

Mailcenter

Calender Features

Conventionel Functionality

Discussion boards

Urgent-Reques

Blackboards

E-Mail Listserver

Member List

synchronous Communication

Member-oriented Facilities

Management-oriented Facilities

PotentialFunctionality

Monitoring of (social) Group Structure

Monitoring of (social) Group Activities

Topic management (Portfolio)

Report-generation

...

Protocol-Data

Extended Functionality

Search Feature

Document Storage

Evaluation Systems

Buddylists

Mailcenter

Calender Features

Conventionel Functionality

Discussion boards

Urgent-Reques

Blackboards

E-Mail Listserver

Member List

synchronous Communication

138 M.Trier

To arrive with a more systematic picture of this multi-channel issue, a first distinc-

tion can be drawn by the differentiation between offline and online channels of

communication, where the latter are employing IT as a carrier for their communi-

cation. The channels of a knowledge network can further be differentiated into de-

centralized versus central networks. A decentralized network has no perceptible

central hub, like a central network. Such a hub could for instance be a web-portal,

where all communication is centered and visible to all. Usually decentralized net-

works include more decentralized communication between two or more people in

the network, without the option, that others can easily participate. These two di-

mensions result in four different types of communication channels:

• decentralized offline networks (e.g. talks between two employees, which

can not be accessed by others),

• central offline networks (e.g. a meeting of all participants at one location,

to which all persons need to travel to),

• decentralized online networks (e.g. e-mail communication, which is

again only visible to the directly participating persons), and

• central online networks (e.g. discussion boards using a software plat-

form).

These four areas differ in their network properties. For example, a decentralized

network like an e-mail network has a different visibility of its member’s contribu-

tions as a discussion board hosted on a central portal. Figure 42 assembles the dif-

ferent types of channels in a matrix.

Figure 42: Box Model of Communication Channels for CoPs.

offline online

central

decentral

+ costs of transaction-

Talk/Dialog

Meeting

Letter

Circular

E-Mail

-Distribution+

Discussion

Board

offline online

central

decentral

+ costs of transaction-

Talk/Dialog

Meeting

Letter

Circular

E-Mail

-Distribution+

Discussion

Board

Visualization of virtual Knowledge Communities 139

The properties of the two different categories offline versus online communication

and their implication for communication and community is being analyzed by

academia using the term Computer-mediated Communication, or simply CMC

(e.g. Sproull and Kiesler, 1986; Riva and Galimberti, 1997). The underlying re-

search discipline approaches to compare advantages versus disadvantages to sub-

stantiate and identify the role and the fields of application for the new means of

communication.

The advantages of computer-mediated communication and their application for

group work are illustrated in Figure 43.

Teams are not limited to meet at specified days but can establish permanent expert

groups which work on different topics. The members do not have to work co-

located but can be geographically dispersed. This is especially important for inter-

departmental lateral communication in companies. Further, it is very unlikely, that

all experts exist at one place. Another advantage is that members can asynchro-

nously discuss their problems exactly at the time, they need the solution. They do

not have to wait until the next meeting. Whereas physical meetings often show

random or unsystematic discussions or a series of inflexible presentations, discus-

sion in a CMC environment can be systematized in threads, which allows parallel

meaningful dialog in topic areas. This enables larger groups, as sub-networks can

dynamically be created. Larger groups in turn improve the access to expertise as

the likelihood is far bigger to meet relevant experts as in a group of twenty. The

communication is protocolled automatically and remains persistent and accessible

in communication archives, together with relevant related working documents. Fi-

nally, multiple persons literally can speak simultaneously. The written form of

discourse allows for following more comments as in an oral setting in a physical

meeting. Even in a chat, the reader can scroll back to read the parts of recent con-

tributions, which are of interest.

Generally, CMC supports the establishment of lateral communication among em-

ployees of the same rank. This eventually is also leading to a more personal type

of communication. The examples in chapter 5.2 show, that many multinational

companies primarily base their communities on such CMC channels in order to

utilize these benefits.

140 M.Trier

Figure 43: Advantages of Computer-mediated Group Work.

However, CMC research is also concerned about the richness and sociability of

electronic media. As current research on CMC is primarily motivated by social

sciences, it concentrates on analyzing the effects of CMC on social relationships

and collaboration in organizations (Riva and Galimberti, 1997). The main ap-

proach is comparing groups which rely on face-to-face communication with

groups employing CMC. Here, researchers added a further differentiation between

synchronous and asynchronous CMC and found that especially in asynchronous

CMC, two typical features of face-to-face conversation are absent (Mantovani,

1996): the collaborative commitment of participants and the co-formulation of the

message and the feedback which allows the social meaning of the message to be

processed immediately. This raises the general question, if CMC lacks necessary

relational features (social cues) which are needed by persons to correctly interpret

their social setting (Sproull and Kiesler, 1986). This social vacuum negatively af-

fects the personal identities of subjects (Sproull and Kiesler, 1991). Obviously, the

missing non-verbal feedback in CMC renders social processes more important

than in a face-to-face situation. Lea and Spears (1991) therefore demand social

reference norms to regulate behavior.

From this difficulty to socialize another difference emerges: CMC shows more of

a task orientation than face-to-face meetings, as it requires a larger effort to social-

ize over CMC (Walther and Burgoon, 1992:62). This aspect is offset by negative

effects in finding consensus in a group. This happens, because there are no social

cues to adhere to and leaders cannot take charge of the discussion without disturb-

ing the system of voluntarism. Further, the increased degree of anonymity reduces

sub-ordinance in the group, so that more people will maintain and communicate

Threaded and systematical

dialog in topic areas

ÆRandom discussions or

inflexible presentations

Multiple persons can speak

at once or read

ÆSequence of talkers, others

listen, or 1-to-1 exchange

Groups can be much larger –

(sub-)networks emerge

ÆClear size limits to ensure

effectivity

Geographically spreaded,

asynchronous, on-demand

ÆCo-located, time-dependent,

not on-demand

Computer

Computer-

-mediated group

mediated group

Physical group work

Persistent and accessible

communication archives with

central related documents

Manual protocols or

inaccessible dialogues,

circling documents

Moderation, Management of

virtual Knowledge Work,

Organizational Integration

Moderation, Management of

Knowledge Work,

Organizational Integration

Permanent expert groups

which work on topics

Teams which meet once in a

while

Threaded and systematical

dialog in topic areas

ÆRandom discussions or

inflexible presentations

Multiple persons can speak

at once or read

ÆSequence of talkers, others

listen, or 1-to-1 exchange

Groups can be much larger –

(sub-)networks emerge

ÆClear size limits to ensure

effectivity

Geographically spreaded,

asynchronous, on-demand

ÆCo-located, time-dependent,

not on-demand

Computer

Computer-

-mediated group

mediated group

Physical group work

Persistent and accessible

communication archives with

central related documents

Manual protocols or

inaccessible dialogues,

circling documents

Moderation, Management of

virtual Knowledge Work,

Organizational Integration

Moderation, Management of

Knowledge Work,

Organizational Integration

Permanent expert groups

which work on topics

Teams which meet once in a

while

Visualization of virtual Knowledge Communities 141

their own opinions (Walther and Burgoon, 1992:52). The decisions that are made

can thus show a tendency to be less compromising (Spears and Lea, 1994:448).

Discussing the issue of CMC and its effects, Knowledge Work in people networks

also directs the attention to a special type of community which has not yet been

mentioned in chapter 4.2 on the typologies of CoPs. In the past years, the increas-

ing application of CoPs in global enterprises together with the event of sophisti-

cated means of global Computer-mediated Communication has facilitated the de-

velopment of CoPs whose members are not co-located (Lesser and Storck, 2001).

Such a community is referred to as ‘Virtual Community’, if people primarily util-

ize a large array of traditional media like phone, teleconference, or fax, and elec-

tronics media like e-mail, videoconferencing, newsgroups, or intranets to support

its members’ interactions.

The term ‚Virtual Community’ has first been used in 1993 by Howard Rheingold.

His positive experiences from the internet-based portal ‘The well’ in the 1980s

and 1990s influenced his sociological conception of the Internet as an utopist

world which is existing in contrast to the local communities in the real world. He

thus defined virtual groups as democratic and equal coalitions of individuals,

which effectively cooperate in a joint venture (Rheingold, 1993). A virtual Com-

munity of Practice is often also called distributed (Wenger et al., 2002), computer-

mediated (Etzioni & Etzioni, 1999), on-line (Cothrel & Williams, 1999), or elec-

tronic (McLure-Wasko & Faraj, 2000). However, it has been mentioned, that often

CoPs employ a mix of communication channels. Thus a Virtual Community can

also meet physically in face-to-face meetings. Conducted on a regular basis, they

have been shown to be important for building relationships and trust among oth-

erwise geographically distributed participants of Virtual Communities (Storck &

Hill, 2000). Still, factors such as geographical dispersion and busy schedules make

virtual (and also asynchronous) communicating through IT more efficient.

Summarizing, the research on CMC and on Virtual Community is showing both,

promoters and skeptics, so that no clear answer can be given to the question, if a

Virtual Community is a proper environment of effective Knowledge Work in a

(social) people network. However, an empirical study of virtual groups (cf. Berge

and Collins, 2000) supports the promotion of Virtual Communities in enterprises.

According to the survey, 72.9 percent of its moderators considered their group as a

community. Further, 70.6 of the moderators believed that their members feel

themselves as part of a community. More than 70 percent where also noting, that

they actively promote the sense of being a community.

These results are also relevant for the approach followed in this book. Here, it has

to be recognized, that for the research objective of conceptualizing support for the

methodical analysis and evaluation of Communities of Practice, the two offline

sectors are very difficult to include. They would require time consuming manual

interviews or surveys with the community members, a field for a conventional so-

cial network researcher. Although such analysis is sometimes done in practice, it

142 M.Trier

would be difficult to apply it for a continuous and near real-time monitoring and

management approach. Quite contrary, the two sectors of online communication

channels are lending themselves more to automated analysis and evaluation. As

this book is written from a business informatics perspective with the objective of

identifying new application fields for Information Technology in a corporate envi-

ronment, the research focus and scope will now be narrowed down to focus the

two areas of online communication. However, the underlying assumption is that

these two channels are to some extend representative for the community activities.

This in turn depends on their weighting in the overall mix of communication me-

dia. Here, according to an analyst prediction, by 2007 the individual’s time spent

interacting with others in the virtual world will exceed physical interactions by a

factor of 10 to 1 (McWilliam, 2000).

6.4 Focusing Discussion Groups

The last chapter highlighted the importance and benefits of Computer-mediated

Communication and its application to Virtual Communities. The large share of

such electronic discourse especially in the application of CoPs in large enterprises

with multiple sites raises the question, which of these channels are the most im-

portant ones. For the conceptualization of a support for moderators and managers

of CoPs (more precisely of CoPs with a large extend of virtual communication),

the abundance of communication channels needs to be broken down into a very

restricted sample of CMC means. To aid this decision, Figure 44 shows the results

of a survey, which identified the most frequented CMC-channels in communities.

On the top three positions, there are Discussion Boards, E-Mail, and Instant Mes-

saging.

Figure 44: Ranking of utilized Communication Channels in two types of Communities.

Source: Ambrozek and Cothrel (2004)

Customer Communities Employee CommunitiesCustomer Communities Employee Communities

Visualization of virtual Knowledge Communities 143

These three top channels will comprise the focus of the subsequent concepts.

However, while including both, central and decentralized online communication

(cf. Section 6.3), the subsequent discussion and design of a prototypical solution

for an improved IT-support will adopt the domain of central online communica-

tion as its primary perspective40. A focus on central online communication is si-

multaneously putting the most frequented communication tool into the center of

analysis: the discussion group (also called bulletin board, discussion group, or

newsgroup). Communication via discussion groups is considered a research chal-

lenge because it is still insufficiently examined and the current interface is merely

text-based. This form allows for a central and topic oriented storage of messages

between experts.

Compared to this means of information exchange, currently much research is done

with e-mail networks. It produced first applicable approaches for an automated

visualization (but not yet true evaluation) of e-mail networks. They capture com-

munication networks between senders and receivers in order to allow for structural

analysis. For example, HP Research has identified 66 communities by capturing

the e-mail communication of its 367 research affiliates. The groups have an aver-

age size of 8 people. Moreover community leaders can be identified (cf. Tyler et

al., 2003).However, these networks have the disadvantage of being a decentralized

peer-to-peer communication concept, where it is very likely to not oversee the

overall content within the network. Quite contrary, discussion groups provide a

consistent and complete access to the insights stored in it. The content is organized

in topic threads. This makes discussion groups a suitable tool for targeted conver-

sation generating conclusions or integrated perspectives. Examples are the devel-

opment of an XML-extension to a web-based programming language, the devel-

opment of an integrated design of a new business process, or the management of

product problems. In all these scenarios, there are requests for expert advice in

sub-domains within a larger topic area. The moderator is responsible for giving

orientation and maintaining momentum within the discussing group.

On the other hand, current discussion boards are not very ergonomic. They pro-

vide features like the generation of threads. One member initiates a posting and

others can reply to it. Over time a tree-like structure of comments forms around an

initial question in a topic area. In larger boards, there can be thousands of semi-

structured text elements posted by many hundreds of people. This makes it quite

difficult to quickly work into the group’s structure or to identify the most impor-

tant areas and most important experts. In large groups, like the general discussions

dealing with the Microsoft Operating System, the size is causing redundant contri-

40 Including both means, that while presenting the subsequent approaches using electronic

discussions as the primary perspective, the architecture of the software solution also

allows for analyzing decentralized networks. The main underlying technical concept

is the connector and its translation of decentralized communication archives into the

generic database, described in chapter 8.3.3.

144 M.Trier

butions, so that constant analysis of the board has been implemented to identify

large overlaps and cross-postings (e.g. Smith and Fiore, 2001). The main reason

for such inefficiencies can be seen in the user interface, which has not much

changed since the first introduction of discussion board technology. Another inef-

ficiency has been discussed by Chang et al. (2002):

“Threading and cross-indexing are useful for organizing newsgroup articles, but are

not so helpful in identifying mutual interactions among the authors. Often a news-

group user is looking for meaningful dialogs among the participants. Threading arti-

cles is not good enough as not necessarily all articles in a thread contribute meaning-

fully to the discussion: There certainly exist ‘me too’ articles and advertisement.

Cross-indexing the articles by author names or subject keywords does not help much

either in finding mutual interaction, as it loses the temporal dimension in group dis-

cussions.” Chang et al. (2002:751)

Obviously, looking at online discussions, the notion of visual components (which

will be discussed in the context of the concept of social translucence in the next

chapter) can also be applied to improve the experts’ communication network:

Oliver et al. (1998) find that interactive materials are essential in a virtual envi-

ronment, as opposed to pure text-based scaffolding. Further, Johnson frames the

question: Can Communities of Practice in their true definition be set up, main-

tained, and supported using current web-based applications, which are mainly

text-based environments (Johnson, 2001)?

Following this research direction, this contribution now examines how the value

creation in electronic discussions of communities can be analyzed by automati-

cally extracting and visualizing useful and already existing data about the commu-

nity structure, consisting of the entities employees, topics, and documents as well

as their many relationships.

In this context, another advantage is, that the analysis of discussion groups does

not cause a privacy problem like with e-mail networks, because the information

contained in it is meant to be public to the members of the group. This public visi-

bility of contributions also causes less ‘noise’ in the messages. This means, that in

a professional application, there are almost no unrelated messages, distorting the

overall conversation.

All these issues render discussion groups a focal communication channel for fur-

ther examination, analysis and visualization of the exchange of knowledge in ex-

pert communities. The main objective is to make online discussions more trans-

parent and hence easier manageable. Only then, the previously introduced re-

quirement of regularly observing and monitoring the work of a Community of

Practice becomes feasible. However, a subsequent future research challenge in

analyzing communities will be the question, how to accommodate and re-integrate

the spread of communication over various channels in the analysis of people net-

works.

Visualization of virtual Knowledge Communities 145

6.5 Current Gaps and required Concepts

The increased practical employment of virtual communication in expert groups

poses new challenges for community managers and moderators. The abundance of

communication channels, the very large group size, and the underlying voluntary

participation with the resulting absence of the ability to execute hierarchical au-

thority results in a complex area of management activities. Especially the role of

measurement has been identified as being important for enabling successful inter-

vention into people networks. The section on current IT applications also shows

that current software offerings primarily target the user group of members and do

not supply sophisticated functionality which supports management. Further, the

applications tend to concentrate on content oriented features. The very important

social domain and also underlying processes of management are, despite some ex-

ceptions, largely ignored.

The results of this situation can be illustrated with a recent study of Ambrozek and

Cothrel (2004). Although 79 percent of moderators and members of CoPs agree,

that technologies for online communities are continuing to improve and participa-

tion in online communities is growing (82 percent), most organizations can’t

measure return on investment (72 percent agree). To a large extend, this is attrib-

utable to the fact, that the discipline of creating and managing communities is

poorly defined (59 percent agree). This results in a situation, where less than half

of the respondents feel, that executives understand the value of online communi-

ties. Here, a systematical approach for capturing, measuring, evaluating and visu-

alizing existing electronic expert networks can be a suitable method to improve

current management approaches. The high value of according IT applications for

such a monitoring of communities has not yet been discovered - neither by soft-

ware vendors nor in industry applications of CoPs. Rather, companies are cur-

rently often only conducting survey-based audits to assess their communities,

foregoing the rich data they could derive from their software and ignoring network

metrics and models to improve the effectiveness of the knowledge network. Using

questionnaires, the current conditions and outputs of the groups are estimated

(Heinold, 1999). The available data about the virtual communication is not used

and integrated into this measurement approach. Figure 45 illustrates current man-

ual approaches. On the top left, the survey procedure is illustrated, including

workshops, the problems of low return rate, report generation and an executive

summary. The actual survey is depicted at the bottom-left side. It mainly includes

issues of satisfaction, which do not really convey structural properties. However,

the bottom-right side shows the achieved evaluation. The example identifies, that

negative experiences dominate: Next to dissatisfaction with the reputation of the

group and problems with its integration into daily work, as a special problem has

been stated, that moderation and facilitation failed.

146 M.Trier

Figure 45: Current manual Approaches of evaluating CoPs.

Summarizing, a demand for augmenting community software with features for

automated monitoring of community activities, social structures and outputs can

be recognized. The main underlying question is how to generate necessary trans-

parency, which in turn relates to the issues of measurement and evaluation. The

next chapter will thus develop an appropriate measurement concept. Its advantage

is that it derives its primary measures from sociological research in order to rec-

ognize the predominant role of a complex (social) communication network be-

tween people. The subsequent chapter will then derive the appropriate extension

of IT-support and further presents several case studies to illustrate the added

value.

Moderator

Workshop Improvement

Answering

Percentage Member-

Survey Report

Generation Result

Discussion

Background

Information Letter

Customization

Incentives result-

database deskriptive

detailreport Executive

Summary

Moderator

Workshop Improvement

Answering

Percentage Member-

Survey Report

Generation Result

Discussion

Background

Information Letter

Customization

Incentives result-

database deskriptive

detailreport Executive

Summary -1

-2,0 -1,5 -1,0 -0,5 0,0 0,5 1,0 1,5 2,0

fail excellen

goodpoor satisfying

Management Support

Daily-Work Integration

Moderation & Facilitation

Marketing & Reputation

Competences

Contributions

IT Infrastructure

Face-to-Face Communication

Sharing Culture / Sharing Potential

Objectives

Overall Satisfaction

Business Value

Support

Structure

Content

I&C

Structure

General fit

Output

Visualization of virtual Knowledge Communities 147

7 Towards CoP Transparency, Measurement and

Evaluation

In the previous section, a gap in technically utilizing the CoP’s rich body of com-

munication data for improving community work and community management was

identified. This section will now develop appropriate methods to provide the re-

quired insights into the actual structure of the knowledge workers and their activi-

ties using their electronic communication data as a source. For that, the first issue

is to establish a perspective on a Virtual Community, which takes into account the

key findings of all previous chapters. Among them,

• the main objective to develop and share knowledge and the support of re-

lated knowledge processes,

• the relationships between people, their organizational affiliation in proc-

esses, their contributions (documents) and topics (according to the KM

entity model),

• the forming of topic related groups or clusters of experts,

• the resulting establishment of a dense communication network with many

channels for informal information flows which extend the hierarchical

formal architecture,

• the different network roles in such communication clusters, and

• the ability for moderators, members, external stakeholders or analysts to

actually observe, analyze and document all these aspects.

These additional aspects will enhance the content oriented work and methodically

complement the current log analysis to improve the effectiveness of utilizing vir-

tual CoPs as Knowledge Management instruments. The two major issues and ob-

jectives for the development of such an extension are namely transparency and

evaluation. Both require an underlying set of factors (in the form of measures or

model elements) to capture the actual community in an appropriate model, which

then can be analyzed.

7.1 Measurement and Success Factors for Knowledge

Networks

A method to establish a measurement system which aids in the facilitation of

Communities of Practice is the initial identification of general success factors.

148 M.Trier

These factors provide the ‘strategic’ background, for which then, on an opera-

tional level, concrete measures have to be assembled.

Research on success factors is generally aiming at formulating guidelines, that can

be influenced by persons and which leads them to an (intervention) strategy that is

likely to be successful (Trommsdorff, 1990). The resulting concepts are usually

not fully explaining all the correlations between the factors, but still they can help

to intervene more effectively into a system (Leihmeister et al., 2004:2).

Following this approach, in this section, the contributions of Vestal (2003),

McDermott (2000), Schoen (2000), Leihmeister et al. (2004), and Roberts (1998)

are introduced. All these authors analyze necessary factors to create successful

Communities of Practice. Their findings provides the basis for the main domains

of measurement and a first set of concrete related factors for the objective of util-

izing electronically available data to better manage Knowledge Communities.

They are assembled to guide the selection of useful measurement domains and in-

dividual measures, which in the end are synthesized into a special measurement

system for virtual Communities of Practice. It has to be noted, though, that not all

success factors are lending themselves to help with the creation of a measurement

system. For example, some of them simply relate to special managerial interven-

tions by management or the establishment of an appropriate IT infrastructure.

A first selection of rather broad factors to be considered in order to move towards

a successful Community of Practice has been proposed by Vestal (2003). His ten

traits for creating a successful community are shown in Figure 46. Although this

list is too general to be directly applicable for deriving concrete measures for some

measurement system, it highlights several broad issues, which substantiate the im-

portance of increased transparency and evaluation. For example, trait 3 suggests

transparency about the core content by providing a knowledge map. This directly

relates to the analysis of subject matters. Trait 4 directs the attention to the notion

of eliciting, measuring, and presenting knowledge sharing processes. Trait 8 de-

mands the provision of a measurement system, which is capable of providing in-

sights about the community’s progress to achieve business results (without actu-

ally discussing the factors that should be included, though). A further very general

requirement is put forward in trait 9: it should be possible to recognize the mem-

ber’s expertise. Related to the last aspect is a factor discussed one year later by

Lesser and Cothrel (2001). They emphasize the development of connections to

people with relevant expertise as the biggest challenge in an online community.

Visualization of virtual Knowledge Communities 149

Figure 46: Ten Traits for a successful Community of Practice. Source: Vestal (2003)

A further general overview about community success factors has been published

by McDermott (2000). He divides factors into four groups: a management chal-

lenge, a community challenge, a technical challenge, and a personal challenge (cf.

Figure 47). Among the relevant issues for a measurement approach are the topic

focus of the CoP, the encouragement to participate, the notion of thought leader-

ship, personal relationships among community members, the activity of the core

1. A compelling, clear business value proposition for all involved

• What value does belonging and participating in the CoP have for an individual?

• What value does it bring my department if one of my staff takes time to participate?

2. A dedicated, skilled leader

• Does the CoP leader have the skills to facilitate an organic, outside-of-line responsi-

bility group?

• Does the leader have a vision for moving the CoP forward?

3. A coherent, comprehensive knowledge map for the core content of the CoP

• Does the group call on frequently used common content, topics or knowledge that

should be pulled into one shared space?

• Do all members of the CoP understand who the sources and recipients of knowledge

are within the community?

4. An outlined, easy-to-follow knowledge sharing process

• Do people know how, what, and when to share?

• Are community members able to easily access and reuse knowledge from others or a

shared space?

5. A technology medium that facilitates knowledge exchange, retrieval and collaboration

• Does it include a repository of community content and knowledge?

• Is the technology supported by the organization’s IT group?

• Does the technology meet user group needs? Did they have input into the look, feel,

and content?

6. Communication and training plans for those outside of the CoP

• Do community members (and prospective members) understand why they should

participate?

• Is there a self-training or short program that shows individuals how to share and find

knowledge from each other?

7. An updated, dynamic roster of CoP members

• Are CoP members able to access others with their interests quickly and easily?

• Do members have tools that assist with rapid, one-to-many communication?

8. Several key metrics of success to show business results

• Does the CoP have a documented measurement system to show how it’s meeting its

business value proposition?

• Is there a plan for collecting, reviewing, sharing and validating measures of success?

9. A recognition plan for participants

• Can participants recognize what’s in it for them?

• Is the recognition scheme built into the HR process and is it part of the development

or evaluation process?

10. An agenda of topics to cover for the first three to six months of existence

• Do your CoP leaders and members have hot problems to solve early in the lifecycle?

• Are there sufficient face-to-face or voice-to-voice meetings for members within six

months of the group’s launch?

• Are there enough actions and activities for this group to become accustomed to

working together to solve problems?

150 M.Trier

group, the relation between knowledge development („thinking together“) and

knowledge exchange („sharing information“).

Figure 47: Critical Success Factors in Building Community. Source: McDermott (2000)

Next to these two American approaches, there are two European perspectives on

the required set of success factors. First, Schoen (2000:220) proposes a very com-

prehensive and detailed list of nine success factors, each including a set of detailed

aspects:

1. knowledge carriers (experts),

2. behavior, culture, and management support,

3. vision, expectations, objectives, and strategy,

4. external networks of the community,

5. IT-infrastructure and tools,

6. CoP organization and structures,

7. CoP processes and activities,

8. contents and context, as well as

9. results and satisfaction.

Again, not all these factors include items which help to utilize available electronic

data to derive useful measures for evaluating CoPs. Still, various important meas-

urement domains can be identified. Taking the perspective of this book’s approach

to look in more detail at the first factor: knowledge carriers, aspects like the sub-

ject fields (knowledge) of the experts, the internal network structure of knowledge

Management Challenge

1. Focus on topics important to the business and community members.

2. Find a well-respected community member to coordinate the community.

3. Make sure people have time and encouragement to participate.

4. Build on the core values of the organization.

Community Challenge

5. Get key thought leaders involved.

6. Build personal relationships among community members.

7. Develop an active passionate core group.

8. Create forums for thinking together as well as systems for sharing information.

Technical Challenge

9. Make it easy to contribute and access the community‘s knowledge and practices.

Personal Challenge

10. Create real dialogue about cutting edge issues.

Visualization of virtual Knowledge Communities 151

carriers with their interests, their similarities and network roles, sub-groups, or the

network’s size and dynamics are of interest. Factor 2 contains aspects like qual-

ity, trust in the network, constructive feedback culture, experimental knowledge

development, or participation activity. Factor 3 subsumes issues like identity or

progress towards objectives. This is also substantiated by Lave and Wenger (1991)

who highlight that it is vital to communicate the community’s objective and pro-

gress. Factor 4 emphasizes the visibility of the CoP to the outside. Other factors

which can help to determine measures are Factor 7 proposing to measure time of

participation, activity level, interaction frequency, or Factor 8, which is among

other things directing attention to the quality of the contributions, the amount of

contents, and the age or recency of a contribution. These factors can help to give

orientation for setting up an own measurement system as described in section

7.1.5.

A further approach to systematizing elements which influence the success of a

Community of Practice has been proposed by Leihmeister et al. (2004). The

strength of this approach is its empirical foundation. The authors observed 160

communities and collected 745 datasets.41 The success factors have been ranked

and are shown in Figure 48. Elements, which are relevant for the developed meas-

urement approach are (in decreasing order of importance) reaction times, the

quality of the contents, recency of contributions, controlling the satisfaction of

CoP members, encouraging interactions among members, the relationships be-

tween old and new members, a continuous evolution of the community which

reflects the interests of the members, trust among members, controlling the

growth of the community’s membership, and controlling the frequency of visits.

A final analysis of important success factors which shall be recognized for the es-

tablishment of the CoP measurement system has been suggested by Roberts

(1998). She found, that the sense of community in a virtual group does not primar-

ily depend on synchronous interaction facilities but on the richness of asynchro-

nous participation opportunities. They can be determined by factors like cohesion,

effectiveness, help, relationships, language, and self regulation. Especially in-

teresting is the factor cohesion, which looks at the relation between long time

members and new users as a proxy for stability, which is directly indicating the

sense of community in a virtual group.

Analyzing the multiple approaches to establish influential success factors for

Communities of Practice can help to provide a broader segregation of the relevant

domains for developing a measurement system as described in section 7.1.5. The

following general key domains can be defined:

41 Unfortunately the sample includes various types of commercial communities like gaming

communities, lifestyle communities, etc. which more belong to the type of Commu-

nity of Interest. However, insufficient research is available to conclude that this leads

to inapplicable results for the field of knowledge communities.

152 M.Trier

• knowledge (in terms of quality contents and according topics), which is

stored within the network together with the according processes to de-

velop and share it, as well as the ability to recognize and access the ap-

propriate experts,

• the participation activities of the members which are necessary to create

dense networks between experts together with the underlying require-

ment of high degrees of trust, and

• a continuous evolution, progress and growth of the community, which

has to be visible to the outside.

Figure 48: Success factors of Virtual Communities. Source: Leihmeister et al. (2004)

Next to these three domains, the issue of group identity is an important dimension.

However, it can not be very well measured using the electronic data available.

Rather, it is implicitly affected by measurement and visualization and hence is in-

Visualization of virtual Knowledge Communities 153

cluded as an indirect and qualitative domain complementing the measurement sys-

tem.

In summary, the conceptualizations of success factors, which have been intro-

duced so far, already suggested various concrete measurement factors for a meas-

urement system:

• (number of) general relationships among community members,

• relation between knowledge development and knowledge transfer,

• activity of the core group (in terms of volume and frequency of participa-

tion),

• important subject fields,

• network size,

• network dynamics,

• reaction times (as indicator of feedback culture),

• (number of) relationships between old and new members (cohesion),

• network growth, and

• frequency of interaction or participation.

Next to the introduction of general success oriented management factors for run-

ning CoPs, an analysis of related research topics is now being introduced. In prin-

ciple, these fields can be derived by taking the above lists of success factors and

the manager’s tasks discussed in section 5.7 as an orientation. After looking at Co-

threl’s (2000) approach of measuring community structures, the emphasis will be

on general group research, empirical social factors, measures proposed by the field

of Social Network Analysis, and finally the field of Social Capital in an organiza-

tion. As the next sections will show, together these four fields motivate various

measures, which can be derived from data of electronic communication and which

can then be used to support management interventions into CoPs. Because of these

qualifying aspects they are incorporated into the integrated measurement system

for evaluating Communities of Practice introduced in section 7.1.5.

7.1.1 Measures from logging Community Activity

A first potential source for useful CoP measures is the work of Cothrel (2000). He

describes a framework consisting of three dimensions: activity, topic, and eco-

nomic measurement. Activity measures describe the health, topic measures show

insight, and economic measures indicate the Return on Investment (ROI).

154 M.Trier

Figure 49: Cothrel’s three Dimensions of Community Measurement.

Source: Cothrel (2000)

Cothrel’s most elaborated domain is the Activity measurement. He suggests a

(somewhat arbitrary) set of ‘common measures’ for an online community (also

compare Figure 50). They include unique visitors, page views, session time,

community click-through (what percentage of visitors to the home page click

through to a community program), registered members, postings per

day/week/month (in member-to-member interaction programs), read-to-post ratio

(in member-to-member interaction programs), page additions (in member-

generated content programs), page revisions (in member-generated content pro-

grams), peak number of concurrent users (in live events), total number of users (in

live events), audience penetration (if the total size of the target population is

known), repeat visits, and frequent visitors.

Figure 50: Common measures for measuring the success of an e-community.

Source: Cothrel (2000)

The second domain is topic measurement. Here, Cothrel remains very general and

simply states that topic measures are difficult to define and that they require con-

tent analysis. He indicates issues like the identification and analysis of heavily

trafficked community areas, discussion threads and the most commonly discussed

subjects. Further, he mentions the analysis of the concentration of interest, de-

scribing how member activity is distributed across the most commonly discussed

subjects. Another factor is the percentage of total community traffic that flows to

the ten most heavily trafficked topic areas and the spread of community interest

Economic

ROI

Health

Insight

Topic

Activity

Visualization of virtual Knowledge Communities 155

across the top ten subjects. Finally there is the share of the top subject in terms of

overall communication volume. Although Cothrel is not giving clear and operative

examples of how to derive these figures, these issues can later be utilized to derive

measures.

The final set of required CoP measures suggested by Cothrel is evaluating ROI.

He understands this community-related ROI as follows:

“Contrary to popular belief, community ROI is not about `monetizing' community

members or performing other unnatural-sounding acts. […] [I]t is about putting a

process in place for recognizing the value that online community members create.''

(Cothrel, 2000:19)

This domain is even less detailed than the very general topic measurement do-

main. Cothrel notes, that this domain requires data not only from the community,

but also from transactions systems in order to allow for a comparison of commu-

nity member activities versus the activities of other site visitors. He is only very

general about the ‘how’ of measuring ROI. Cothrel suggests metrics like visits,

return visits, purchases, frequency of purchase, transaction size, and referrals.

Here, Cothrel’s focus on commercial communities has fundamental differences to

the approach of this book. He assumes that the conversion of community members

to paying customers generates the ROI for the community. However, this is not

the case in Knowledge Communities. Here the generated value is less a consump-

tive one but more an increase in overall efficiency. Obviously, the applicability of

Cothrel’s suggestions for the economic domain is limited to commercial customer

(B2C) communities, showing again the importance of determining and defining

the appropriate types of the community under consideration (compare section 4.2).

Summarizing, Cothrel’s topic domain is not systematized and needs to be trans-

formed into a clearly defined set of measures. He does not discuss the underlying

motivations and relations of his measures (e.g. to more general structures of

CoPs). The suggested ROI measures can not be applied for the scope of this

book’s measurement system. However, Cothrel’s collection of activity measures

can be of help for establishing a basic quantitative impression. Among the impor-

tant ‘common measures‘ are:

• read-to-post ratio,

• peak number of concurrent users,

• registered members, and

• postings per day/week/month.

Unfortunately, the main focus lies on listing the typical logging measures for

tracking the usage of web-based communication applications. Nevertheless, this is

an important domain for measuring the activity of a Virtual Community although

it has to be noted, that is insufficient if not augmented by other measurement do-

mains. This aspect is indicated in the following quote:

156 M.Trier

„Using hits and page views to judge a site's success is like evaluating a musical per-

formance by its volume.“ (Forrester Research, 1999)

Hence, for the purpose of deriving meaningful measures, pure statistical activity

numbers like e.g. total number of users online are not providing useful insights.

They need to be connected to an interpretational framework. The parallel users

could for instance be interpreted as an indicator of co-presence which in turn im-

proves the feeling to participate in a living group. To closer examine this interpre-

tational framework, more abstract research areas, which are related to the commu-

nity’s structure and processes need to be analyzed.

7.1.2 Measures from Group Theory Research

As Communities of Practice can be regarded as a social group, an important con-

tribution to the research question of how Communities of Practice should be struc-

tured, run, improved, and supported can be found by reviewing general group the-

ory. This can then help to derive the factors to be measured in order to manage the

community as suggested by these theories.

Although this field is far too broad to be extensively discussed within the scope of

this book, the factors which are relevant for analyzing and supporting Communi-

ties of Practice are introduced now.

As a first approach to define the term group using its basic properties, Wiese

(1924), identifies duration and continuity, role structure, and emerging tradi-

tions. This rough set of group properties constitutes a first concrete set of general

factors which can be analyzed in order to be able to assess if group properties are

existing as a part of evaluating qualitative properties of a community. Especially

the group’s duration can be analyzed although it has to be added, that there is no

concrete point in time defined for a set of persons to become a group. However, in

the other direction, if some stability and duration can be observed it is a positive

indicator of a Community of Practice (although it should be enriched by analyzing

other factors like activity).

Schaefers (1980:20) demands, that a successful group requires a system of shared

norms and a role structure. The shared norms emerge during interaction which is

affected by social sanctions for not behaving according to the established norms.

If feedback signals successful communication behavior, than this behavior is re-

peated and manifested. Others may eventually imitate these patterns. The role

structure hence emerges from these patterns and from the activities of members.

For example, if a person often organizes, delegates, and speaks, it is likely to take

a leading role. If it is a specialist in a special subject field, it will be asked on that

topic and develop it further.

A further general factor for the persistence of a group is interaction, although its

type and intensity can vary between groups. Only by interaction - which in a com-

munity setting is meaning communication - norms, structures, and identity can

Visualization of virtual Knowledge Communities 157

emerge. Festiger (1950) describes four causal factors for communication in a

group:

• communicating opinions and values,

• achieving objectives of the group,

• changing or forming roles and status, and

• conveying emotions and perceptions.

Communication can also be analyzed according to its contents (cf. chapter 8.4.3

on content coding) and according to its directions. The latter is independent of the

content and conveys information about the factor ‘role structures’. Direction is

recognizable, when the most active persons are asked often. Further, when lower

positions direct their requests to individual persons and not to the whole group, or

if the overall communication flows from lower to higher ranks in the network.

The communication structure can also be related to the performance (i.e. suc-

cess) of a group. For example, tasks are solved best in a star-pattern of communi-

cation. Lines and circles are usually slower and more apt to mistakes.

Lerner and Becker (1962) have analyzed the impact of similarity or homogeneity

between members. They conclude, that in competitive situations complementary

persons collaborate, whereas in situations where shared activities are executed,

homogeneous persons build groups. Homogeneity as such ascertains each partici-

pant in its own behaviors and values and increases stability in a group. In a Com-

munity of Practice, where the transfer and conjoint generation of new knowledge

is the primary objective, homogeneity between members is hence an important

factor. However, it should also be noted, that if people are too similar in their

knowledge and values, it is very likely that their similar perspectives will decrease

their ability to innovate (in terms of integrating different domains and perspec-

tives). A greater variety of members creates a better pool of resources, which leads

to higher performance (also compare the section on Social Capital).

Summarizing, the main factors which can be applied for generating an interpreta-

tional framework for evaluating statistical measures in a virtual CoP are:

• duration,

• continuity,

• role structure,

• emerging traditions / shared norms,

• interaction (in a certain frequency about contents in several direction),

• communication structure, and

• homogeneity.

158 M.Trier

7.1.3 Measures from Social Network Analysis

Social Network Analysis (SNA) is defined as a framework for the analysis of

structured social relationships (Wasserman & Faust, 1994), which in the organiza-

tional context can reflect role-based authority relationships of formal organiza-

tional structures, informal structures based on communication, information ex-

change, or affection (Tichy et al., 1979).

The main hypothesis of this field of research is that human behavior is influenced

by structural properties (e.g. restrictions) and less by direct characteristics of the

actors (e.g. norms). This shifts the focus away from classifying actors using their

properties towards observing relationships between actors and the actor’s em-

beddedness in a complex relationship network. This relationship network and the

individual relationships have influential structural conditions. Hence SNA is gen-

erating and applying approaches and instruments to describe and evaluate the rela-

tionships between actors. This focus is very suitable for the analysis of Knowledge

Communities, which consist of networks of people which communicate with each

other. Here, according to Wellman (1997), social networks are virtually present

whenever a group of people interacts electronically. Thus, assuming such commu-

nities are resting on an underlying social network enables the systematic examina-

tion of such computer-networked communities (Wellman et al., 1996). The rela-

tionships between them are of value for the community and the organization. For

this object of analysis, the rapid and regular advance in social networks research

provides a vast body of measurements and methodologies (Wasserman & Faust,

1994), which can also be applied to increase the transparency of people networks

in Communities of Practice and to evaluate their properties. It further provides a

concrete set of measurements which can be used to indicate and clarify the broad

factors proposed in group theory.

In theory, a social network is a social relational system with the elements actors

and relations (relational ties). Actors can also be analyzed on different group lev-

els, like dyads (pairs of actors in two states, adjacent or not adjacent), triads (three

nodes subgraph with four states: 0,1,2,3 connections), subgroups, groups, or the

overall network (Wassermann and Faust, 1994).

There are two general and related types of networks. An ego-centric network con-

sists of a focal actor, termed ego, a set of alters who have ties with ego, and the

measurement on the ties among ego and these alters. This data is also referred to

as personal network data. Ego networks are used frequently to study the social

support of persons (Wasserman & Faust, 1994). Each alter in an ego network has

an own personal network again, so that finally all ego networks interlock to form a

complete social network of a group.

Measurements within the domain of SNA can either include composition vari-

ables, i.e. the properties of actors, or structural variables, i.e. the properties of rela-

tionships. These relationships can be directed or undirected. In the first case, the

Visualization of virtual Knowledge Communities 159

direction of communication is unidirectional, e.g. A supports B. Undirected rela-

tionships represent connections without a specific direction, e.g. A and B commu-

nicate. However, this relationship still can be asymmetrical (e.g. B writes more

than A). Further, relationships in a network can differ in their strength. Usually a

person has few strong relationships and various weak relationships to peers, al-

though there is no specific definition of what is weak and what is strong. Gener-

ally, this property is relative to the general constellation in a social network

(Marsden and Campbell, 1984). It has been found, that pairs who maintain strong

ties are more likely to share the resources they have (Wellman and Wortley,

1990). Another important finding for the analysis of expert networks in the con-

text of Knowledge Management has been provided by Hansen (1999). He found,

that strong ties are important for sharing tacit knowledge. Further, an electronic tie

combined with an organizational tie is sufficient to allow the flow of information

between people who may never have met face-to-face (Garton et al., 1997).

The applied sociological approach of sociomatrices derives various network prop-

erties from a very simple set of data about the network stored in a matrix. The

rows and columns each belong to a person. The value in the cell, where row A

meets column B represents the strength of relationship between both persons A

and B. This relationship could for example be the amount of communication ex-

changed between two employees.

The simplest measure is to actually count the number of members or actors in a

network. It is a proxy for the network size. Larger social networks tend to have

more heterogeneity in their social characteristics and more complexity in their

structure (Wellman and Potter, 1997). Large heterogeneous networks (such as

those often found online) are good for obtaining new resources whereas small ho-

mogeneous networks are good for conserving existing resources (Garton et al.,

1997).

A further indicator for network size is called diameter. It is the longest distance

between two nodes in a network or in other words, the longest path. A small line-

shaped network of 6 people (longest path or diameter = 5) is thus larger as a dense

network with several hundreds of people around one central node (here the diame-

ter is only 2). It is a proxy for the likelihood that information passes through the

net. The larger the diameter, the less likely is this information flow.

Further, the actually-occurring relationships between the actors in a network can

be counted. They can be compared to the number of possible ties. This is calcu-

lated by multiplying the size of the network’s population (n authors) by (n-1) and

dividing this by 2 if the tie is undirected. A network of 6 people can hence have

(6x5/2=) 15 ties if undirected relationships have been under consideration. The

network measure density is now the ratio between existing ties and the maximum

possible ties. A density of 1 means all possible ties exist. Hence density indicates

if a network is tightly or very sparsely connected.

160 M.Trier

“Densely-knit networks (i.e., groups) have considerable direct communication among

all members […] few members of sparsely-knit networks communicate directly and

frequently with each other. As in the Internet, sparsely-knit networks provide people

with considerable room to act autonomously and to switch between relationships.”

(Garton et al., 1997)

The density can be applied together with relationship strength to discover clusters

or groups in a network. They usually are highly interconnected sets of actors (thus

employing a high density). Further they have strong ties within the group. Such

evaluation of subgroups or epicenters of communication can help to identify struc-

tural patterns of the overall network. All measures can also be applied to compare

and characterize the subgroups. Such network groups can be connected to other

network groups by actors with membership in both groups.

Another basic property of an actor is his nodal degree. This is the number of lines

that are incident with it, which is equivalently the number of adjacent nodes. It is a

measure of the activity and connectedness, assuming that connections need activi-

ties to maintain. A network can be characterized by its average nodal degree,

which shows the average number of connections of each actor. If an actor has a

degree of zero, i.e. no direct relationships, he is called an isolate.

In people networks with directed relationships, it can be differentiated between

indegree (e.g. A has received support) and outdegree (e.g. A provided support).

Applying these two indicators to Communities of Practice can show, who is very

active (i.e. sending messages or having a large outdegree) and who has much at-

tention or prominence in the network (i.e. large indegree by receiving messages

from others). Outdegree is hence a measure of expansiveness whereas indegree is

a proxy for receptiveness and popularity (Wasserman and Faust, 1994:126). Com-

paring the indegree and outdegree of two actors in a relation gives an insight

about their relationship’s symmetry, which can also be named reciprocity (see

next chapter for its interpretation in the social domain). If an actor has an indegree

of zero and an outdegree larger than zero, it is called transmitter, the opposite con-

stellation is named receiver. If both types of degrees are above zero it is called a

carrier (Wasserman and Faust, 1994:128). In the context of the measure degree,

there are also the network positions hub and pulsetaker. A hub is a person with

many direct ties. It is very likely to simultaneously be a broker. Such a broker is

connecting large subsets, which would otherwise not be connected. A pulsetaker is

having only few direct links to the network, which makes his position very com-

fortable as every link requires effort to be maintained. Still, he has very effective

links, for example to a hub, meaning that while only having few direct links he

can draw resources from the network through a multitude of indirect links, which

originate in his connected actors.

Related to the nodal degree is the measure of centrality. It is an important prop-

erty of a network as a very centralized network is dominated by few or even just

one central node. If such critical nodes are removed, the network fragments into

Visualization of virtual Knowledge Communities 161

unconnected sub-networks. Networks with low centrality have no single point of

failure.

One subtype of centrality is degree centrality, which is simply indicating, how

involved the actor is in ties (i.e. having a high degree; Wasserman and Faust,

1994:178). A second subtype is closeness centrality, which measures how close

an actor is to all other actors in the network. If he is close to all, he can quickly

interact with them without needing to rely on many other people (Wasserman and

Faust, 1994:183). A third type of centrality is called betweenness centrality. It

assumes that an actor, which is required to be passed if two other actors want to

connect, is ‘between’ these two actors. The more instances of being ‘between’ two

actors exist, the higher is the ‘betweenness’ of the actor under consideration. It is

in other words simply the number of paths that pass through a node and indicates

how often a given node is required by others to reach any node via the shortest

path. Actors with a high betweenness are central in the network as they are under-

going a certain ‘stress’ during the activity of the network. The measure between-

ness can be normalized by dividing it by its maximum value.

The notion of betweenness also relates to the network role of a broker. Such a

broker occupies an important network position. It sits on an exclusive path be-

tween many other actors. They hence require this person to connect to many of

their related actors. Such a broker between two subnets can also be called a cut-

node. If this node is removed, many connections between other actors disappear.

A related name for this network position is gatekeeper, as in a diffusion process,

this node can control the flow of information (Garton et al., 1997).

In conclusion, this section briefly introduced the perspective of Social Network

Analysis together with its most important measures. These can help to establish

useful insights into the structure of a virtual Knowledge Community network be-

tween experts, which mainly consists of communication acts between the mem-

bers. The most important factors for evaluation with a measurement system are:

• network size,

• relationship strength,

• network roles (broker, gatekeeper, pulsetaker, hub, isolate, transmitter,

receiver, carrier),

• degree (activity, prominence, symmetry or reciprocity),

• betweenness and centrality,

• density, and

• diameter.

162 M.Trier

7.1.4 Measures for Social Capital or Trust

Social Capital and trust have already been introduced as a major benefit and re-

source for a company (Chapter 5.3.2). Measurement and facilitation of community

structures need to include this dimension. That is why this chapter is discussing

the main factors which can be measured to indicate the extent of Social Capital

and trust in a virtual Knowledge Community. It has to be noted that most of these

measurements actually originate in the domain of Social Network Analysis. Social

Capital can hence be regarded as a more abstract property emerging in social net-

works. The likelihood of receiving social support is less dependent on the personal

characteristics of an actor, but more on the form and quality of his relationships.

In chapter 5.3.2 on Social Capital as an organizational benefit, the three interre-

lated structural, relational, and cognitive dimensions have been described (Na-

hapiet and Ghoshal, 1998). Together, they contribute to the formation of Social

Capital. The structural factor requires the formation and actual existence of infor-

mal networks between persons in order to enable them to identify people with re-

quired resources. Here, the structure of people’s relationship networks is of inter-

est as it determines their access to social resources. Applying the Social Network

Analysis methodology introduced in the previous section, this is further deter-

mined by the relationship’s strength. In the context of Social Capital and trust,

‘strong ties’ and ‘weak ties’ both can be useful for different purposes. Whereas

strong ties imply more trust and familiarity, weak ties were found to provide links

to external sources or other clusters which are often a useful source for novel in-

formation or different perspectives of other knowledge domains (Granovetter,

1973). This leads to the conclusion that the pattern and the strength of relation-

ships is also of interest for analyzing this domain of measurement.

The relational requirement of Nahapiet and Ghoshal is supporting the exchange

between the individuals. It addresses issues like trust, shared norms and values,

obligation, expectation and identification. The relational aspect of Social Capital

requires, that people think their actions will be appropriately reciprocated (cf.

Lesser and Prusak, 1999). This implies that mutual trust emerges from mutual ob-

ligations and the reciprocity of communication relationships. The absence of per-

sonal profiling and the support of altruistic and assisting behavior is another sup-

portive group behavior to improve the relational aspect of Social Capital.

The cognitive aspect of Social Capital finally includes issues like common context

and vocabulary, which is supported by the use of common artifacts and stories.

This shows the importance of measuring similarity and homogeneity to enable

familiarity between persons.

Further, social support is determined by emotional support or help in the past, by

relatedness or familiarity, and to some extend by gender as women tend to provide

emotional support more often than men (Wellman and Wortley, 1990).

Visualization of virtual Knowledge Communities 163

Section 5.3.3 highlighted the importance of a frequent and positively attributed

relationship to establish trust between communicating persons. Trust is a funda-

mental principle necessary to create Social Capital in a knowledge network. Its

determinants are very similar to those of Social Capital, so that these two concepts

can be treated as one interrelated domain.

One measurable proxy to determine the level of trust in a network is hence the fre-

quency of interaction between authors and with this also the relationship

strength as more frequent relationships will also have higher relationship

strength.

A recent study showed that the level of trust which exists in virtual workgroups

could be measurably improved by even a single face-to-face interaction at the be-

ginning of the project (Rocco, 1998). However, this success factor is beyond the

scope of what electronic communication can support. A detrimental environment

which limits the building of trust and mutual obligation is a highly competitive,

individualistic culture (Lesser and Cothrel, 2001). That is why factors like concur-

rency and personal profilation have to consider that for a group, the people who

actually help others are the most helpful members. Hence, altruism should be re-

warded.

Another factor which supports trust is mutual obligation (Nahapiet and Ghoshal,

1998). The underlying idea is that an actor is more likely to provide resources on

request, if he has an obligation to do so. Such an obligation emerges, when he has

received more resources from the requesting actor, than he has provided. This

concept is thus analog to paying off a debt.

Jarvenpaa and Leidner (1998) analyze the forming of trust in global virtual

groups. They identify several behaviors which influence trust. Positive early be-

haviors include social communication, conversation conveying enthusiasm, in-

dividual initiative, and conjointly coping with technical and task uncertainty. All

these factors can be measured to some extend by conducting a manual or semi-

automatic content coding (also compare section 8.4.2 on content analysis). This

method classifies different types of communication and thus allows for analyzing

the shares and networks of different such categories. Behaviors, which try to im-

prove trust in more mature stages of the group, include predictable communication

(e.g. warning of absences), timely responses, rotating leadership, and phlegmatic

reaction to temporary crises.

In summary, the domain of Social Capital research obviously draws on measure-

ments from Social Network Analysis. It serves as an abstract layer or an interpre-

tational framework above the quantitative measures of SNA. This will later also

be reflected in the hierarchical dependencies between different domains of meas-

urement in the measurement concept. As a result, the most important factors

which can be supportive for an integrated measurement system for Knowledge

Communities are:

164 M.Trier

• homogeneity,

• reciprocity or mutual obligation,

• familiarity,

• relationship structure and strength,

• interaction frequency,

• altruism and assistance to others,

• degree of social communication, and

• response time.

7.1.5 Measures for Knowledge Processes

A final domain, which has no underlying theory, but can be derived from the other

areas, is the measurement of knowledge processes. However, assuming, that the

Community of Practice conducts Knowledge Work in a network of experts, this

domain is implicitly evaluated by the other domains. Given, that the knowledge

processes are contained in the actual communication, it has to be analyzed, to

what extent the actual transfer of knowledge can be identified by communication

analysis.

Main questions are the relation between questions and answers, or the time until

responses are available. Another important related domain is the actual analysis of

topics, which have been discussed. This can be done by extracting keywords or

analyzing broad and related topics.

Summarizing, it has to be noted, that no theoretical frameworks exist so far for the

evaluation of actual knowledge processes in expert communication. Although a

set of measures will be developed to reflect this domain, more research is still to

be done to be able to automatically evaluate knowledge transfer or knowledge de-

velopment. The main factors so far are:

• relation between questions and answers,

• response time, and

• topic analysis, including similarities or dominating topics.

7.2 Deriving a Measurement Concept for Communities of

Practice

After having analyzed the main tasks of management (section 5.7), the underlying

success factors for running a Community of Practice (section 7.1), and factors in

five related research areas, a large and substantiated selection of important meas-

Visualization of virtual Knowledge Communities 165

urement factors or indicators can be assembled. The objective remains to identify

available electronic data in virtual Knowledge Communities which can be utilized

to calculate the discussed measures and to integrate them into a systematic meas-

urement system. This can then be used to direct the design of a supporting soft-

ware tool for analyzing virtual communication networks (see the next part of this

book on the development of the according software tool).

Figure 51: Measurement Concept for the Evaluation of virtual Knowledge Communities.

Figure 51 shows the conceptualized measurement approach for evaluating CoPs.

The multiple measures identified in the previous sections can be systematically

classified into three types:

1A) measures of structure (quantitative communication structure including basic

logging data),

1B) measures of Social Capital (Social Network) (including group factors, social

network factors, and more abstract extend of social capital and trust), and

1C) measures of knowledge processes (knowledge development and transfer).

These three domains constitute the core of the measurement system. The simplest

layer is the actual quantitative structure. It can be evaluated by looking at meas-

ures like number of authors, volume of messages, average time between messages

sent, network density, network diameter, etc. The structural measurements can

also be calculated for each relation or for each author. A more general indicator of

this domain is structural uniformity, which implies the evenness of the network. It

detects steep peaks or very evenly distributed network activity.

Structure

Social Capital

(Social Network)

BKnowledge

Processes

(Developm. &Transfer)

113

Measuring improves

the perception of

identity

Provides base

measures for

examination of growth

Visualizations &

Measures Visualizations &

Measures

Growth

Content Analysis

Identity

2 3

166 M.Trier

Some structural properties also convey information about the social network

within the communication network and by this also about the quality of Social

Capital as the access to resources distributed in a people network. This section

thus comprises a second layer on top of the quantitative structure. Examples are

average strength of relationships, reciprocity of relations, number of direct and in-

direct contacts per author, etc. More complex soft indicators which can be evalu-

ated with such measures are the level of trust and the degree of Social Capital.

Some of the structural properties additionally give insights about the level of

knowledge exchange and its underlying knowledge processes. Useful measures

are unanswered versus answered questions, average number of replies, or average

message length per relation. This knowledge related set of measurable network

properties also leads to content oriented analysis of the network structure. In the

measurement concept, this is the point, where topic and keyword analysis are inte-

grated into Social Network Analysis in order to align the idea of analyzing com-

munication networks with the idea of Knowledge and Expertise Management42.

Measures are e.g. the main keywords of a relation, an author, or the complete net-

work as well as the similarity between authors based on their keyword overlap.

Next to these three core sets of measurement two related domains are included.

Identity (domain 2) is the first important domain. Here, facilitation of virtual com-

munication networks can be achieved by simply enriching the otherwise invisible

group structures with useful visualizations. Examples are cues of who is similar to

an ego, adjacent, or co-present. This phenomenon of Social Translucence has been

discussed by Erickson and Kellogg (2000) and will be provided as a brief intro-

duction in the next section.

Next to supporting group identity, the three structural properties finally provide

the foundation form the basis for the final suggested measurement domain: growth

and development (domain 3). It helps to understand not only the configuration of

the network, but also its dynamic behavior including the group’s velocity or de-

celeration, its declining sectors or the establishment of network roles over time.

To prepare for this novel domain of measurement and evaluation of dynamic be-

havior, a longitudinal visualization component has been developed. It allows for

actually observing how new nodes and relations are added over time and how

network properties are changing. In the future, this element provides much poten-

tial for further research as such analytical insights about longitudinal network

measures are of interest for community moderation and management, especially if

it is connected to topic analysis.

The complete set of individual measures for the four domains Structural Proper-

ties, Social Capital Properties, Knowledge Process Properties, and Growth Proper-

42 More on this innovative yet very challenging endeavor can be found in chapter 8.4.4.

Visualization of virtual Knowledge Communities 167

ties is shown in Figure 52. All measured elements are related to the overall net-

work (marked with N), to individual authors (A) or to relations (R).

Together the table includes 55 individual measures for a Virtual Community. They

can be differentiated into 13 structural properties, 18 Social Capital properties, 16

knowledge process properties, and 8 growth properties. The amount of measures

already implies the focus on analyzing social elements and knowledge processes.

Some of these measures are now introduced in more detail to illustrate the general

approach of deriving the evaluation indicators from the available electronic data.

Structural measure number 5 is called ATBMS. This is simply the abbreviated

form for ‘average time between messages sent’, which is a proxy for the frequency

of postings and also implies the speed of content additions or content related

growth. It can be calculated for the network and for each author. For the network

(notation: N [ATBMS]) it is simply the overall lifetime of captured data (last –

first entry) divided by the sum of all messages (N [Duration] / N [#msg]).

Structural measure 11 is the core group’s share of the overall number of authors

in the network (N [Coregroup share]). To determine this core group, the most ac-

tive actors in an expert network are ranked. Then they are added to the core group

until this group accounts for 80 percent of the network traffic. The share is then

the size of the core group divided by the total number of available actors in the

network (N [Coregroup size/ # A]). This measure shows, if the core group is either

very small compared to the rest, showing that there is a ‘tall peak’ in the commu-

nication volume with only a small but very active nucleus, or if the network is

more evenly distributed. It is thus similar to a market share or monopoly measure

in economical theories.

Social Capital measure 14 indicates an author’s obligation in an individual rela-

tionship (A [Obligation in R]). It indicates how many an author ‘gives to’ or ‘takes

from’ a contact. Obligation occurs, when the author has generated a ‘debt’ in

terms of support, meaning that he received more from his peer than he provided. It

has been shown in section 7.1.4, that this increases the probability, that he pro-

vides support at the next request of his contact person. The measure simply relates

the received messages to the sent messages and normalizes this number to arrive

at a range of -100 to +100 percent of obligation (A [1-((#sent M - #rec M)/ (rec M

+ #sent M))-100]). Reaching the maximum 100 percent means that the author has

a very high obligation to provide resources at the next request of B.

Social Capital measure 18 is the similarity of author properties and requires au-

thor coding for its actual implementation (R [Property Similarity of A's]). This

means that it is necessary to analyze a wide range of information about the au-

thor’s properties. The measure basically calculates a similarity measure by count-

ing similar properties (R [#pairs of property entries/# properties*100]). This could

for example be organization, division, hierarchy, job role, years of experience,

prior projects, industry sector, education, etc. The theoretical idea is that a high

168 M.Trier

similarity between authors (in terms of backgrounds) positively affects the levels

of trust and subsequently also the access to other employees’ resources.

An example of a network role analysis is Social Capital measure number 24, the

pulsetaker. The measure compares a person’s indirect contacts with its direct con-

tacts and computes the relation between these two factors (A [#indir contacts /

#direct contacts]) in order to identify how many indirect contacts the actor has per

direct contact. This helps to identify pulsetakers, which have to put very little ef-

fort into the maintenance of their direct contacts and still have access to many in-

direct sources of information.

Measure 28 is a very easy to compute yet very powerful indicator for the net-

work’s probability of providing social support: the average strength of the rela-

tionships of the network (N [avg strength of R]). It is calculated by counting and

averaging the numbers of messages for each relationship between two authors (N

[avg R [#msg)]). The stronger the relation in the network, the higher is the prob-

ability to access other member’s resources.

A further interesting example is measure 31, the number of mutual contacts in a

relationship between two authors (R [# mutual contacts]). The underlying argu-

ment is that the probability for a good relationship (and hence for access to re-

sources in the network) increases when the two persons share some contacts (R [#

paired contacts of A and A2]).

Measure 32 looks at the amount of relations between old and new members of an

expert network (N [#R [old and new]]). It is calculated by looking at the overall

maturity (time period) of the expert network and counts relations which connect

two authors, whose days of first appearance in the network are very distant to each

other - at least half the age of the overall network (N [# R [where ((days since A

appeared - days since A2 appeared)/days of discourse data) > def50). This is a

very interesting proxy for the likelihood of new members to connect and draw on

resources of experts.

Measure 35 is an example for a measure for evaluating knowledge processes. It is

comparing answered versus unanswered postings in a virtual discussion (N [an-

swered/unanswered msg]). This can technically be done by looking for messages

which reference another message. These are then regarded as answers (N [#initial

M with reference / #initial M without being referenced by others]). It is a proxy of

how likely it is to receive answers for requests or in other words to receive support

from experts.

Visualization of virtual Knowledge Communities 169

Figure 52: Four Domains Model of Network Evaluation

(Structure, Social Capital, Knowledge Processes, and Growth).

Description/Name Calculation Implies

1 1 N [Message Volume] N [#msg] Size of the network in terms of content

2 2 N [Author Volume] N [#A] Size of the network in terms of participants

3 3 N [Relations Volume] N [#R] Amount of relations in the network

4 4 N [Diameter] N [longest path] Size of the network in terms of distances

5 5 N [ATBMS] N [N [Duration] / N [#msg] ] Frequency of activity, speed of added contents,avg time between mess sent

6 6 N [#Transmitters] N [# A[no received M] Only sent messages, similar to minus 100 of A [Obligation]

7 7 N [#Receivers] N [# A[no sent M] Only received messages, similar to plus 100 of A [Obligation]

8 8 N [Variance of #contacts] N [Variance of #contacts] Structural differences in the network (steep)

10 10 N [Variance of Relstrength] N [Variance of Relstrength] Structural differences in the network (steep)

11 11 N [Coregroup size] N [# A[coregroup=y] Size of active nucleus

12 12 N [Coregroup share] N [Coregroup size/ # A] Share of active nucleus in overall activity

13 13 A

[

ela

sed time since last a

A [days since last message sent or received] How long has the author been idle/inactive?

14 1 A [Obligation in R] A [1-((#sent M - #rec M)/(rec M + #sent M))-100] (-100% to +100%) How many authors gives or takes to a contact

15 2 R [Reciprocity] R [(1-((abs(#rec M - #sent M))/#(rec M + #sent M)))*100] approaches 100% when symmetric relationship

16 3 N [avg Reciprocity] N [sum reciprocity / # R] shows how even the relationships are

17 4 A [avg Obligation] A [sum A[Obligation in R] / #R] how much obligation has an author to others, negative for accessing network ressources

18 5 R

[

Pro

ert

Similarit

of A's

]

R [#pairs of property entries/# properties*100] [0..100] - same background (like hierarchy), more similarity more trust and access

19 6 N [Density] N [R (existing) / R max * 100] Connectedness in the network, = access to all ressources

20 7 N [# Isolates] N [# A with 0 R] no established or historical access to network ressources

21 8 N [Density of Coregroup] N [R (existing) / R max * 100 for all R between A[coregroup]] how well connected is the nucleus

22 9 A [# core group contacts] A [R where A2 [# core group = y]] how many active and central access persons

23 10 A [direct contacts] A [#R] how many access persons

24 11 A [Pulsetaker] A [#indir contacts / #direct contacts] how many indirect contacts per direct contact (multiplier of indirect information access)

25 12 A [Broker] A [# R [where A is neccesary path-element of R[A2,A3]]] how many pathes between other authors in the net do require A as their element, bottlene

26 13 N [#Brokers or Cutpoints] N [#A where A [# R [where A is neccesary path-element of R[A2,A3]]] how many weak points are in the network structure

27 14 A [duration of membership] A [(days since A appeared/days of discourse data)*100] is it an established member or a recent

28 15 N [avg strength of R] N [avg R[#msg)] how strong are on average the messages in the network - good for access to ressources

29 16 A [avg strength of R / #R] A [avg R[#msg)/#R] how strong are the relationships of A on average

30 17 A

[

above avera

e connectio

A [#dir contacts/ N[avg A[#dir contacts] is an author extensively connected compared to the avg network

31 18 R [# mutual contacts] R [# contacts of A appear in the contacts of A2] Propability for a good relationship increases when same 'friends'

32 1 N[# R[old and new]] N

[

# R

[

where

((

s since A a

eared - da

s since A2 a

eared

)

/da

s of discourse data

)

how many relationships between oldies and newbies exist

33 2 N [av

mess len

th in chars]N [avg mess length in chars] how likely is content being included in a message how comprehensive is it

34 3 N [Assistance] N [#A with avg obligation <0] or N [# A [obligation <0)/#A *100] how many authors assisted others more than they took

35 4 N

[

answered/unanswered m

N [#initial M with reference / #initial M without being referenced by others] only hierarchical discourses! How well get questions answered

36 5 N [avg #answers/request] N [avg #referencing M for #initial M] only hierarchical discourses! How well get questions answered

37 6 N [avg answer time] N [avg time between M and referencing M (not only initial msg] only hierarchical discourses! How well get questions answered

38 7 A [avg answer time] N [avg time between M and A's referencing M (not only initial msg] only hierarchical discourses! How well get questions answered

39 8 A [Rating] A [avg rating by others] requires property: Rating, quality measure

40 9 R [Keyword Similarity of A's] R [#pairs of keywords/# keyowrds * 100] [0..100] - same background (topic-related) - Content Analysis based similarity measure

41 10 N [keyword list] N [20 most frequent keywords identified - together with their neighboar words] Content Analysis - shows content of the network

42 11 A [keyword list] A [20 most frequent keywords identified - together with their neighboar words] Content Analysis - shows content of the authors contributions

43 12 R [keyword list] R [20 most frequent keywords identified - together with their neighboar words] Content Analysis - shows content of the relation

44 13 A [keyword variance] A [#M containing 1 out of 20 most frequent keywords/#M ] Content Analysis - How many of A's messages are covered by his 20 keywords - polarizati

45 14 N [keyword variance] N [#M containing 1 out of 20 most frequent keywords/#M ] Content Analysis - How many messages are covered by 20 keywords - polarization

46 15 A [topic leaders] A [most occurences of the networks keywords] Content Analysis - who speaks most about the keywords

47 16 A

[

Network ke

word overla

A [how many overlapping keywords with N[keyword list] Content Analysis - similarity between the author and the overal network

48 1 N [Density Trend] N [Density change over time] how improves the connectedness between authors

49 2 N [Sliding Density Trend] N [Density of relations idle for < 20% of the discourses period - change over time] how improves the recent connectedness between authors

50 3 N [Diameter Trend] N [Diameter change over time] how improves the size of the network over time in terms of distance

51 4 N [ATBMS Trend] N [N [Duration] / N [#msg] - change over time] how increases the speed of message volume addition in the network

52 5 A [ATBMS Trend] A [A [Duration] / N [#msg] - change over time] how increases the speed of message volume addition of the author

53 6 N [Ke

word Variance over ti

N [#M containing 1 out of 20 most frequent keywords/#M - change over time] how changed the polarization of the network over time

54 7 N [Hot Topics] N [Keywords, whose occurances increased most over time] which keywords spread across the network over time

55 8 A [Hot Topics] A [Keywords, whose occurances increased most over time] which keywords determined A's contributions over time

Structural Propert

Social Capital Propert

Knowledge Process Propert

Growth Property

170 M.Trier

Another knowledge process evaluation is provided by measure 40 – a keyword

similarity measure between two authors (R [Keyword Similarity of A's]). It is

simply looking for the amount of keywords, which two selected authors have in

common to derive an insight about their shared topics of expertise. This is then

being normalized to arrive at a percentage which represents the amount of overlap

of the author’s backgrounds (R [#pairs of keywords/# keywords * 100]).

In the domain of growth measures, measure 48 is the trend of the network’s den-

sity. Density has been defined as the actually occurring relationships in a network

divided by the maximum possible relationships. A trend indicator now shows how

the network’s connections do increase or decay over time (N [Density Trend]; also

compare section 7.1.3). Increased density indicates that authors establish more di-

rect connections to each other, thus increasing their own connectedness to others,

which in turn makes a smooth flow of information more likely. Further, it benefits

the social structure as more relationships imply that more paths for social support

are emerging. This growth measure is hence also related to the Social Capital and

the knowledge process domain.

A final measure, which shall be introduced, is measure 54: the identification of

‘hot topics’ in the overall network (N [Hot Topics]). It is actually an enumeration

of topics and not an ordinal or cardinal measure. However, it is of high interest for

moderators in the field. It helps them to actually see, which topics are growing. It

is a growth measure as it is looking at the keywords and their occurrences over

time. The keywords, which increased in their occurrence have spread most across

the network over time and are hence of high interest (N [Keywords, whose occur-

rences increased most over time]).

These dozen measures, which have been introduced in more detail, show the mo-

tivation and the method for deriving a set of concrete indicators by using the

available electronic data. In a subsequent step, such a measurement system can

also be supported with useful visualizations of the results. In the domain of virtual

group visualization, this supports the final domain of identity, which is heavily

relying on mutual awareness of the group and its properties.

7.3 Social Translucence - adding Visualization to

Measurement

Next to mere measurement of statistical factors, the visual access and interpreta-

tion of such factors is a major issue. It is a very beneficial aspect of a software

supported measurement system and increases manageability by providing cockpit-

like features to moderators and members. This is why visualizations of expert

networks are a key element of the evaluation and analysis approach of this book.

Sometimes these visualizations yield more insights than mere numbers as they di-

rectly and user-friendly represent structures, clusters, dead areas, important ex-

perts, dense relationships, etc. However, they need to be connected to measures in

Visualization of virtual Knowledge Communities 171

order to allow for comparisons between different communities. A further benefit

of including visualizations in the evaluation of Knowledge Communities is the

very positive effect on the member’s impression of identity. In the measurement

system indicated in Figure 52 (page 169), this role is reflected by adding a com-

plete domain ‘identity’ which basically hosts visualizations and is a result from

increased transparency through measures.

In this context, visualization can be defined as the use of computer supported, in-

teractive, visual representations of abstract data to amplify cognition (Card et al.,

1999:1).

In the literature, the importance of visualizations for the identity of a group has

been analyzed under the name Social Translucence. This concept has been sug-

gested by Erickson and Kellogg (2000) to express the necessity of providing visi-

ble clues based on perceptual information of the social situation (presence and ac-

tivities of users) to the members of a virtual group to create social resources that

help to structure the online interactions. By such visual information, people be-

come aware of each other and social conventions and dynamics are enriched. The

mutual awareness increases the accountability of the member’s actions. Behavior

is more visible and persistent, the history and thus also the character of users is

getting conveyed. This creates more coherent, productive, and fluid online interac-

tions. Users can imitate and observe others, peer pressure emerges. Erickson and

Kellogg see Social Translucence as a fundamental requirement for supporting

most types of communication and collaboration in digital spaces.

The general underlying benefits of visualization have also been discussed by Card

et al. (1999:16). The authors propose six mayor ways by which an Information

Visualization can amplify cognition:

• Increasing memory resources available to the user: This can be done

by either directly using the resources of the visual system or help cogni-

tion by reducing working memory requirements for doing a task by al-

lowing to store certain information externally and visually. For example

when doing a subtraction by hand, working memory requirements can be

reduced by using a pen and paper and writing down the overlap of the

subtraction of two digits under the next pair.

• Reduced search: Search time can be reduced by visually grouping re-

lated data. Larger amounts of data can be represented in a small area. For

example in a map height of locations is often color coded. If the data

would be printed out for each location it would be probably illegible.

• Enhanced recognition of patterns: Recognizing information is often

easier then recalling the information. Information can be simplified and

organized by grouping certain information, abstracting the information in

an informative way and selectively omitting certain details. UML dia-

grams are used in programming to visualize the structure of objects in a

172 M.Trier

program. They put the different objects in relation to each other through

grouping and thus reduce the complexity of the code by only displaying

an abstraction of its functionality.

• Perceptional inference: Visualization can help people to make a large

amount of graphical inferences which else would be very hard to derive.

It can also enable humans to make specialized graphical computations,

like using diagrams to generate hypotheses about data.

• Perceptional monitoring: By organizing a display, a large number of

potential events can be monitored. This is used in cockpits of airplanes.

• Manipulable medium: A configurable medium allows the user to ex-

plore the parameter space and the data. For example, math programs al-

low the user to plot data and then to zoom into sub areas. The user can

change parameters and directly see the changes in the graphical output.

These general advantages show how beneficial the addition of useful visualiza-

tions is for interpreting a complex set of measures. In an approach that builds on

modeling and analyzing a community to subsequently apply a measurement sys-

tem with a large set of measures, the addition of various illustrating visualizations

is very synergetic. It helps to increase the social translucence of the system for its

members and gives them a better impression of identity. Further, for the target

group of moderators, the visualizations are especially helpful to aid the under-

standing of the moderator or manager about the complex structure of his network.

The developed visualization environment should hence amplify cognition in many

different ways. It should be a configurable medium and allow the user to zoom

the data to different levels of detail. The medium should allow exploring the data

by filtering using different criteria. It should reduce search for data by plotting

many authors and their relations in a structured way in a small area. The applica-

tion should enhance recognition of patterns. This can be done with a structured

display of the data by making the relation dependencies as clear as possible. The

user should be assisted to make perceptional inferences by visually encoding

relevant information into the visualization.

In summary, this section derived a concrete measurement system with 55 meas-

ures in four domains from a selection of success factors. These main domains are

structural quantitative measures, measures indicating the social structure and So-

cial Capital, measures for analyzing knowledge processes, and measures indicat-

ing growth and evolvement. Further, the importance of adding visualizations to

the measurement system has been emphasized. These visualizations often achieve

more benefits in terms of user acceptance and insights as pure statistical evalua-

tion. This is why a major decision for the development of the prototypical soft-

ware has been the predominance of visualization over measurement and evalua-

tion. Whereas visualization enables understanding, measurement enables compari-

son. (“Visualization comes first, measures are the extension.”). In the next section,

Visualization of virtual Knowledge Communities 173

the developed technical software approach to implement the introduced measure-

ment and visualization approach for monitoring virtual communities is described

in detail.

174 M.Trier

8 A Tool for visualizing, analyzing and modeling

Communication Networks of Knowledge

Communities

The analysis of Communities of Practice and their most important elements, prop-

erties, and tasks, the description of the management role, the introduction of the

current state of the art in community-related software applications, and the identi-

fication of important gaps of current CoP platforms highlight the requirement to

provide an add-on to the existing IT solutions, which helps to make Virtual Com-

munities more transparent and which thus also helps to better employ this organ-

izational structure of experts for a network oriented Knowledge Management.

This chapter will now specify the requirements for a tool which provides such

functionality for visualization and evaluation of otherwise invisible expert net-

works. Afterwards, the engineering approaches and decisions of the according

software project are introduced and discussed to finally present the developed so-

lution: a monitoring support for existing software applications for Virtual Com-

munities.

The design stage was guided by a general procedure. Its four main elements are

technical access to data sources, refinement of data to achieve information using a

measurement system, and visualization of the information and provision of a so-

phisticated user interface to match the business objectives and requirements of the

community manager (see Figure 53). The main principle in the development of the

community visualization and evaluation tool is to start from user requirements.

Here, the moderator of a CoP has been selected as the most important role for us-

ing the tool. The moderator puts forward a series of business objectives (block 1 in

Figure 53), which have to be matched with the available data sources (block 2).

For that, the data structures need first to be accessed and captured. The next issue

is to implement useful measurements which bring the original data closer to the

business objectives (block 3). A final component is the visual interface which

links the moderator with the elicited information from the original data (block 4).

The general requirements of a (knowledge) community manager have already

been discussed to a great extend in the previous chapters. Similarly, the measure-

ment system has been conceptually developed in chapter 7.2. Remaining is thus

the examination of the available data sources and their general structures and the

development of a suitable visualization interface, which presents the measures in a

suitable format to achieve insights for the community manager. The latter compo-

nent can be called ‘cockpit’ as it allows a steering person to access and read the

main parameters of a system (just like in a car or plane). This aspect is an ongoing

Visualization of virtual Knowledge Communities 175

field of research as no visual access to virtual people networks yet exists. After

briefly reviewing related technical software approaches and prototypes, the devel-

opment project for this book’s approach will be presented.

The key research areas for the following development of the software for analyz-

ing and visualizing virtual Knowledge Communities are a modeling notation (us-

ing Graph theory and SNA), the depiction of information about network properties

(enriching information by property mapping), the challenge of accessing multiple

channels of communication (and the resulting integration of different storing para-

digms), the actual implementation of the measures for evaluation, the integration

of topics and keyword analysis, and finally longitudinal visualizations, animations

and measures to evaluate evolution and growth.

Figure 53: Overview about the general Procedure to develop a Tool for Visualization and

Evaluation of Communities of Practice.

8.1 Current Visualization Approaches

During the past years, research on Information Visualization proposed some ex-

perimental instruments which employ various visualization techniques, which help

to create a foundation for a management cockpit for social relationship monitor-

ing. Still, they have not yet been combined and applied for the purpose of stake-

holder communication. In order to give an impression about the current state of

the art in related approaches, this section is reviewing the most important proto-

176 M.Trier

types of the field. They usually primarily increase the identity of a virtual group

by adding insightful visualizations of group structures and processes. Most devel-

opments obviously assumed, that the structural patterns are best communicated by

using visual proxies of quantitative community patterns. However, the measure-

ment domain is always inseparably connected to such visualizations as it provides

the basis for generating the individual visualizations. The main strands of meas-

urement and visualization is user logging and its map-like representation on a dis-

play. Only few approaches actually focus on the emerging relationship network of

people in Communities of Practice.

Figure 54: Visualizing Discussion history with the ‘Babble’ Time-line.

Source: Erickson and Laff (2001)

A first example to be introduced is the Babble-Timeline developed by the IBM

T.J. Watson Research Center in order to visualize discussion histories (see Figure

54). This graph is connected to the management domain of social processes. Hori-

zontal lines refer to users, gray parts indicate a user who is logged in but not view-

ing the current discussion, gaps represent periods, where the user was logged out.

Colored parts imply that the user looked at the discussion currently examined.

Vertical marks indicate the user's postings, gray marks stand for postings in other

discussions, and colored marks are contributions to the actual discussion.

A second example is the interface of Coterie, created at the MIT Media Labs

(Figure 55). It is representing users as ovals, their color brightens if they are ac-

tive, they form clusters around topics, active conversants are centered, and inac-

tive listeners (called ‘lurkers’ are located at the sidelines. Despite providing in-

sights in the current movements, the solution lacks a statistical perspective for ex-

post analysis of social processes or structures.

Visualization of virtual Knowledge Communities 177

Figure 55: Display of a Chat Conversation with Coterie. Source: Spiegel (2001)

In 1999, Xiong and Donath created a visualization called Peoplegarden. It repre-

sents users in a message board as petals of a flower-like structure. The height of

the ‘flower’ is related to the time the user is active in the board. There are two col-

ors for the flowers’ petals: magenta and blue. Magenta represents postings, blue

replies. This shows if users only answer others’ request or if they ask more than

they answer. Further the petals start to fade out over time. Thus a discussion group

with bright flowers indicates recent posts. It can be observed if a discussion board

is dominated by one person or if conversations are more evenly distributed. Op-

tionally, the number of responses is represented as small circles at the end of the

posting pattern. Despite the very ergonomic proxy this visualization is not show-

ing the connections between people, so that there are no real networks emerging.

Figure 56: Visualization of Discussion Boards as a Peoplegarden.

Source: Xiong and Donath (1999)

Donath et al. (1999) developed a further visualization which represents postings

over time. It is called Conversation Landscape. Vertical lines show the period of

activity for each participant; horizontal lines are postings. Highlighting shows,

178 M.Trier

who was within hearing range of the selected participant at any given time. This

hearing range is requiring the user to locate himself nearby a discussion in order to

be able to follow its content. If his position is too far away, the user can only see

activities, but can not read the text. The user can interact with the representation,

e.g. by selecting a horizontal bar shows the text of the according posting. The

wider this line, the longer is the related posting. Again, the actual interaction and

knowledge exchange between users is not represented.

Figure 57: Visualizing Author Timelines with Conversation Landscape.

Source: Donath et al. (1999)

Figure 58 shows a completely different approach. Here, the Netscan hierarchy dis-

play (Smith and Fiore, 2001) visualizes the entire Usenet. Newsgroups are repre-

sented by rectangular regions. Boxes are nested within other boxes, indicating hi-

erarchies. Their size represents each group’s cumulative numbers of posts. In the

interactive version, pointing the mouse at a box brings up a text label identifying

the newsgroup or hierarchy. The size can also be set to show the number of post-

ers, the ratio of posts to posters, the average message length, or the percentage

change in message traffic from the previous week to the current one. In a later ver-

sion of this box plot, color coding has been added to indicate shrinking groups

with red colors and growing groups in green.

Visualization of virtual Knowledge Communities 179

Figure 58: Boxplotting the size of different Newsgroups with the Netscan Hierarchy Dis-

play. Source: Smith and Fiore (2001)

Another representation of the Netscan approach shows the actual thread and how

it evolves and differentiates over time (cf. Figure 59). In this view, boxes represent

messages. Lines between messages show replies between messages. Various signs

give more information. Small people-shaped figures at the left hand side represent

the users, which posted on that day. The box with the red dot indicates the origina-

tor of that thread. Thus, it can be observed if this person ever revisited his request.

The half-shaded box finally indicates the most prolific author’s activities.

Figure 59: Netscan Thread Tree. Source: Smith and Fiore (2001)

A somewhat more intuitive display of thread structure and ‘health’ has been de-

veloped with the application eTree (see Figure 60). It is using an ‘ecosystem’ as a

proxy. Threads are mapped as branches with posts as leaves. Recent posts are

colored in a lighter green, eventually turning darker over time. Hot topics are

highlighted as yellow ‘flowers’. Participants are located around the tree (color

180 M.Trier

represents affiliation). The more active a poster is, the closer he moves to the cen-

ter.

Figure 60: Visualizing thread ‘Health’ and Development with eTree.

Source: Girgensohn et al. (2003)

While threaded views on communication like that of Netscan or eTree are useful

for indicating newsgroup action, they are not very helpful to identify mutual inter-

action and the resulting relationships between authors. Thus, meaningful dialogs

between authors are harder to follow (Chang et al. 2002:2). A different approach,

which is more focusing on such relations between clusters of a social network is

the Social Network Browser of Chang et al. (2002) shown in Figure 61.

Figure 61 Browsing Newsgroups with a Social Network Analyzer.

Source: Chang et al. (2002)

Visualization of virtual Knowledge Communities 181

The number of articles of an author determines the nodesize. The prestige degree

(i.e. received messages) is represented by the shade of the node (the darker the

more prestigious). Correspondences between authors are mapped to the edge. This

means that all direct paths (messages) between two authors become one edge with

the according weighting. When there are clusters, where any author has a connec-

tion to all other authors, they are called strongly connected component (SCC) and

are indicated by a box around this cluster.

The tools introduced so far target the user of a virtual communication network.

However, a further target group consists of social network researchers and

analysts. There are software applications offered for this group, which are

concentrating on a very sophisticated set of statistical functionality. Figure 62

shows the application Pajek, which is dominating in terms of spread and

complexity. Despite the multitude of available statistical manipulations, their

meaning is often very difficult to understand without appropriate statistical

knowledge. Further, the application requires a very specific data format which is

not recognizing any special attributes of virtual communication data as Pajek is a

generic network analysis tool.

Figure 62: The user interface of Pajek.43

Summarizing, there are various interesting visualization approaches available.

They reach from simple add-ons to existing community websites to very complex

statistical tools for professional network researchers. The view on social networks

in a Virtual Community environment is still very rare. Rather, visualizations

43 Available as download at: http://vlado.fmf.uni-lj.si/pub/networks/pajek/

182 M.Trier

simply illustrate the flow of events, like for example in thread structures. Some

approaches find metaphors which allow the user to intuitively understand the

development and ‘health’ of the electronic group. Unfortunately, most of the

applications are bound to a specific communication application (or project web

site) and are hence not applicable for arbitrary sets of data. The alternative sector

of statistical tools relies on thoroughly prepared matrices and datasets, which they

import as a special format. Their generic approaches do not allow to incorporate

community-specific information, like author affiliation, or message evaluation.

The requirements of a professional corporate application and hence the

requirements of a moderator of such a group and his need to communicate

progress are not specifically met by current tools.

Despite the partial approaches to provide more insights into social structures and

interaction patterns of communities, there has not yet been a focus on the require-

ments of corporate application.

Hence, even after reviewing a software market with various related tools, it can be

recognized, that there is still a need for developing an application, which over-

comes these shortcomings by allowing to connect to various data sources of vir-

tual communication, which is representing the existing relationships between peo-

ple, and which allows to visualize, manipulate and evaluate the resulting networks

of communicating experts.

8.2 The Tool’s underlying Graphical Visualization Method

The previous examples of visualization approaches emphasized, how measures of

structural community patterns can be enriched by visual proxies. In chapter 7, the

measurement domains of social networks (Social Capital) and knowledge proc-

esses where identified as contributing most to the evaluation of Knowledge Com-

munity structures and processes. For these two areas, useful visualizations need to

be developed. For the prototypical software engineering, the visualization of social

networks has been selected as the primary view.44 It best represents networks of

experts and their interactions resulting in strong relationships. The perspective fur-

thermore elicits meaningful dialogs between experts and is rendering visible the

Knowledge Work which is to be supported by a network oriented Knowledge

Management (also compare for the KM entity model in section 2.2.3). This view

also provides the most sophisticated theory of measurement. To incorporate

knowledge processes into such a view, content and keyword analysis will be

linked to the SNA perspective, so that in subsequent stages of development the

network oriented view can also be extended to represent relationships between

topics or relationships between authors and topics. This novel approach of inte-

grating topic mapping and social network visualization is described in more detail

44 Similar to the approach of Chang et al. (2002), introduced in the previous section.

Visualization of virtual Knowledge Communities 183

in chapter 8.4.2 and 8.4.4. In this section, a brief introduction of the underlying

domain of social network visualization is presented in order to understand the

main elements of the resulting network graphs. These graphs are produced from

the communication data and get laid out using a spatial layout algorithm as devel-

oped in chapter 8.4.1, which also introduces the related technical solutions in

greater detail.

As an extension of the social network analysis using sociomatrices (compare sec-

tion 7.1.3), the related analytical approaches concentrating on network graphs en-

able a more visual analysis of large people networks. This visualization approach

actually originates in the works of Moreno in 1932. He introduced points which

represent actors and edges which indicate the link between actors. The sociomatrix

can be transformed into a graph by taking the matrix and deriving the number of

authors (which equals the number of rows or columns). The graph has the same

number of nodes. In every cell of the sociomatrix, there is a value, which is either

zero or above zero. If the cell has a value larger than zero, the author of that row is

having a relationship with the author of that column. This relationship is added to

the graph by drawing a line from author A’s node to author B’s node. A network

drawing emerges and authors who have many relationships appear well connected.

Figure 63 shows one of the earliest examples of a social network graph. It has

been rendered by Moreno in 1934 and maps friendship choices among fourth

graders. After the invention of these graphs in the 1930s, there where several

stages of development reflecting the respective state of the art in information tech-

nology, like the introduction of computational procedures in the 1950’s (e.g. Bock

and Husain, 1952; Proctor, 1953), first screen oriented graphs in the 1970’s

(Lesniak et al., 1977) or the event of social network analysis tools in the 1990’s

(Krackhardt et al., 1995). A comprehensive overview about this development is

given by Freeman (2000).

The actual graph has not changed since that, still it was possible to automatically

draw very big networks very fast on a screen, to manipulate them according to the

analyst’s needs, and to enter more information into the resulting view. As an ex-

ample, the last section introduced the tool Pajek which can be applied for this task.

Figure 63: Original Drawing of Friendship choices among fourth graders.

Source: Moreno (1934:38), cit. in. Freemann (2000:3)

184 M.Trier

8.3 Software Engineering Process and Software Framework

With the final determination of the main visualization method in the previous sec-

tion (social network visualization) the software engineering process can be

planned and initiated. The main objectives of the development of an application

for visualization and measurement of virtual Knowledge Communities are:

• the connection to various offline and online information sources of vir-

tual communication software (Community Software),

• the provision of a generic and standardized database structure which is

storing different discourses in one structure (e.g. allowing for integra-

tion),

• the development of sophisticated visualizations of networks of experts in

Virtual Communities, and

• the development of a comprehensive measurement system (“visualization

comes first”) to enable the comparison and evaluation of community

structures.

The overall project procedure is shown in Figure 64. Altogether thirteen separate

steps were defined, which form a sequence. However, this sequence is also an it-

erative process as at any time the project can step back, e.g. to include new in-

sights about measurements into the requirement catalogs.

Figure 64: Software Engineering Approach for the Prototype.

Develop-

ment

and

Test

Development

Planning and

Preparation

Systems

Design Systems

Analysis

2. Proof of Concepts – Graphic Algorithms and Visualization Development

5. Digital Mock-up – Graphical User Interface

1. General Requirements – Management, Measurement, and Visualization Factors

3. User Requirements – Configuration of Datasets and their Visualization

4. Technical Requirements – Data Handling Visualization Rendering File Management

6. Database Architecture Design – Tables, Keys, and Attributes

7. Class Architecture Design – Classes, Objects, Inheritance, Communication

8. Setup Developer Environment (IDE)– Eclipse

9. Setup Collaboration Tools – Concurrent Versioning System, Mailing List, Website

10. Definition Project Components and Team Structures

11. Assignment of Requirements to Project Components and Teams

12. Developing required Features and Integration into the Project

13. Cross-testing of Features and feedback from Project Domains, Function‘s Release

Visualization of virtual Knowledge Communities 185

Prior to the actual development work (stage 4: Software development and test),

various preoperational tasks were conducted. First, requirement lists and catalogs

were assembled to reflect general management and measurement, user, and tech-

nical requirements. To check if the electronic communication data actually results

in meaningful networks, various ‘proofs of concepts’ have very early been pro-

grammed in the simpler language PHP45. In stage two, the systems design, a digi-

tal mock-up of the graphical user interface has been generated to discuss user er-

gonomics and the necessary sliders, buttons, panels, etc., using the actual visual

appearance of the tool. Starting from initial circle graphs of social networks, this

process has shown useful for testing the planned computer graphics algorithms.

Architectural graphs of the intended class and communication architecture as well

as the applied database structure have been developed. In the next stage of systems

development planning and preparation, the project time plan has been generated,

the integrated development environment (IDE) has been selected and installed, a

set of collaboration tools has been adopted, the teams were determined and were

assigned to groups of requirements from the requirement catalog. Finally, the de-

velopment stage could start. The required functionality has been generated and

after a testing review and feedback from other teams, it has been released.

8.3.1 Requirements Profile

In stage one of the software engineering project, the main requirements were as-

sembled. These included general requirements for community management and

the according measurement or visualization factors, user requirements of configu-

ration and manipulation of the dataset to achieve an individual visualization, and

the resulting technical requirements for data handling, visualization rendering and

file management.

Among other things, the user should be able to connect to virtually any electronic

communication network, like NNTP46-based standard Newsgroups, Slashdot

based HTML47 Discussions, MBOX48-based Unix E-Mail Archives and

Listservers, Outlook E-Mail Archives, Lotus Discussion Databases etc. This ne-

cessitated the development of a component based individual connector structure.

All data should be incorporated into one single database. This requires a feature

for updating existing discourse imports with new data. The application needs to be

able to extract keywords from discourses and calculate network analysis measures.

When the utilized data set needs to be extended with user-specific background

data (like author affiliation or evaluation), the addition of editable author and rela-

tionship properties needs to be provided. All data should be transformed into a

45 Pre-Hypertext Processor

46 Network News Transfer Protocol, also see chapter 8.3.3

47 Hypertext Markup Language

48 Mailbox Mail Standard for Unix Operating System

186 M.Trier

meaningful visualization graph. This requires renderers for 2D and 3D networks

of communication. Visual manipulation should include rotating and zooming the

resulting network graphs. A feature for time-based observation of growth of the

network has been listed to later enable longitudinal measures and visualizations

which where found to be very useful for network evaluation. The application

should be able to store screenshots and optionally videos from the networks and

their evolvement. Picking of authors needs to be supported in order observe their

individual behavior, development, and author properties. Intelligent features to

reduce the network size are required to increase transparency in big people net-

works, e.g. the limiting of visible authors enables to generate egonets and partial

networks of his immediate environment and contacts.

A sample part of the catalog with the main requirements is given in the following

Figure 65.

Nr Bereich Nutzungsanforderungen, Der Nutzer muss abgeleitete Technikanforderungen, die Software muss Priorität Bearbeitun

sstand

1 Quellenimport ein unterstütztes Daten-Format in die MySQL Datenbank

importieren können im Pull-Down Menu 'Import' die unterstützten Konnektoren

identifizieren (welche existieren) und als Menupunkte anzeigen

2 Quellenimport ein unterstütztes Daten-Format in die MySQL Datenbank

importieren können für lange Quelle>MySQL Transfers eine Fortschrittsanzeige in

Prozent anzei

3 Quellenimport ein unterstütztes Daten-Format in die MySQL Datenbank

importieren können der

ewählte Konnektor identifiziert und auf

erufen werden

4 Quellenimport ein unterstütztes Daten-Format in die MySQL Datenbank

importieren können der entsprechende Konnektor die Auswahl erhalten

5 Analysedurchführung Nach Abschluss der MySQL Aufbereitung wird der Nutzer

gefragt ob die Analyse sofort starten soll oder später eine Dialogbox zur Auslösung oder zum Abbruch der Analyse

anzeigen

nal

sedurchführun

- Lo

die Anal

se verfol

en können während der Anal

se einen Fortschrittsbalken anzei

nal

sedurchführun

- Lo

die Anal

se verfol

en können während der Anal

se lo

texte in einem Fenster anzei

nal

sedurchführun

- Lo

die Anal

se verfol

en können das Anal

selo

in ein lokales lo

file able

9 Analysedurchführung - Log die Analyse verfolgen können die möglichen Visualisierungsformen auf Basis der erhältlichen Daten

eingrenzen (2D-Cluster, 3D-Cluster (evtl. über Pajek VRML

Exportdatei einlesen und in Webfenster anzeigen), Circle Diagramm,

Spring Diagramm, Egonetwork, Dendrogram, Charts, Measurement-

Log,

10 Analysedurchführung - Log die Analyse verfolgen können die nicht möglichen Visualisierungsformen im Menupunkt

Visualisation auf disable stellen

nal

sedurchführun

- Matri

die Anal

se verfol

en können im Hinte

rund die Matrix bilden und in einem Arra

vorhalten

12 Analysedurchführung -Visual. die Analyse verfolgen können im Hintergrund die Visualisierungsformen erstellen und auf

Grafikpanes zu Anzei

e vorhalten

13 Analysedurchführung -I-Menu die Analyse verfolgen können im Hintergrund die zu den Visualisierungen gehörenden

Interaktionsmenues erstellen und vorhalten

nal

sedurchführun

-I-Menu die Anal

se verfol

en können im Interaktionsmenu die Reiter Graph, Lo

und Report anle

15 Analysedurchführung -I-Menu die Analyse verfolgen können im Reiter Graph die Möglichkeiten zur Manipulation der

Visualisierungsdarstellung anzeigen (Nodesize>Activity, Hide

Isolates, Labels: Names, Numbers, Locations, Desymmetrize Matrix,

Size +/-, Fixpoints, Search for terms,…

16 Analysedurchführung -I-Menu die Analyse verfolgen können im Reiter Log die Möglichkeiten zum textuellen Anzeigen der

Netzwerkinhalte aufzeigen (Nutzername, Inhalte der Konversation,

Initiativen, Antworten,…später ausbaubar in Eingriff in das Forum!)

17 Analysedurchführung -I-Menu die Analyse verfolgen können im Reiter Report die Möglichkeiten zur Informationsspeicherung

vorbereiten (Snapshot des Bildframes, Speichern des Snapshots,

Eingabe von Kommentaren zur ggw. Visualisierung in ein

Eingabefenster im Reiter, Anzeige der Reportstruktur in Dialogbox

(Grafik, Autokommentar, Manuellkommentarfelder, Umordnen der

Reportstruktur in Dialogbox, Export in PDF-Datei)…

18 Analysedurchführung - fertig die Analyse verfolgen können die Fertigstellung der Analyse und Vorbereitungsphase in einem

Dialo

fenster anzei

19 Visualisierung eine gewünschte Visualisierung auswählen können ein Pulldown Menu Visualisation mit den angebotenen und für das

aktuelle Datenset möglichen (enabled) Visualisierungsformen

anzeigen

20 Visualisierung eine gewünschte Visualisierung auswählen können die Visualisierungspane und die entsprechenden Reiter werden

ezei

21 Visualisierung einen gespeicherten Snapshot zur Anzeige bringen im lokalen Log-Verzeichnis gespeicherte Grafiken einlesen und

unabhän

vom

ggw

ktuellen Datenset anzei

22 2D-Cluster das 2D Cluster ansehen und manipulieren der Abstand aller Nutzer wird über FR-Algorithmus aus deren

Interaktionsdichte ab

eleitet

23 2D-Cluster das 2D Cluster ansehen und manipulieren Aktivere Nutzer werden mit einer größeren Nodesize kenntlich

emacht

24 2D-Cluster das 2D Cluster ansehen und manipulieren Starke Beziehungen werden über dickere Kanten kenntlich gemacht

25 2D-Cluster das 2D Cluster ansehen und manipulieren die Schwerpunkt der Interaktion wird über einen kleinen grauen

Knoten auf der Kante repräsentiert (Ausgänge/Summe

Ausgänge+Eingänge > der Knoten rutscht zum relativen Empfänger)

26 2D-Cluster das 2D Cluster ansehen und manipulieren das ein

estellte Standardlabel ist der Nutzername

27 Export eine bestehende Matrix abspeichern können einen Menüpunkt 'Export' besitzen

28 Export eine bestehende Matrix abspeichern können ein Dialo

box für ein lokales Dateiauswahlmenü besitzen

29 Export eine bestehende Matrix abspeichern können das Pajek Flatfile Format Exportieren können

30 Export alle möglichen Automatischen Report Elemente automatisch

als PDF exportieren können das Menu Export muss einen Menupunkt für Export des

automatischen Reports anzeigen

31 Datenbank Der Nutzer muss Kategorieinformationen in die Datenbank

eingeben können Die Datenbank muss eine Zusatzspalte bekommen, in der man

Kategorien zu den Messages zuordnen kann low

32 Datenbank Der Nutzer muss Kategorieinformationen in die Datenbank

eingeben können Die Datenbank muss eine Zusatztabelle bekommen, in der man

Kategorien zu den Autoren zuordnen kann (z.B.

Standort,Jobposition)

low

33 Datenbank ein unterstütztes Daten-Format in die MySQL Datenbank

importieren können Die Datenbank muss mindestens die Spalten MessID, From, To

gefüllt haben

34 Konnektoren ein unterstütztes Daten-Format in die MySQL Datenbank

importieren können die Konnektoren zu Newsgroup, PHP-MySQL Archiv DB, Lotus-

Flatfile-Import, Microsoft Messenger aufweisen

35 Konnektoren ein unterstütztes Daten-Format in die MySQL Datenbank

importieren können die Konnektoren müssen ein eigenes Dialogfenster anzeigen

36 Konnektoren ein unterstütztes Daten-Format in die MySQL Datenbank

importieren können die Konnektoren müssen die Eingabe der Quelle erlauben

37 Konnektoren ein unterstütztes Daten-Format in die MySQL Datenbank

importieren können ein HTTP/FTP basiertes Auswahlmenü für das Laden von remote-

locations besitzen

38 Konnektoren ein unterstütztes Daten-Format in die MySQL Datenbank

importieren können ein Eingabefeld für einen Zielnamen für die MySQL Datenbank

Conversion anzeigen, dabei bestehende Datenbankeneinträge zur

Ergänzung anbieten

39 Konnektoren ein unterstütztes Daten-Format in die MySQL Datenbank

importieren können das Dialogfenster des entsprechenden Konnektors anzeigen

40 Konnektoren ein unterstütztes Daten-Format in die MySQL Datenbank

importieren können der entsprechende Konnektor die vorgesehene Quelle finden und in

dessen Daten in die MySQL Datenbank parsen

41 Konnektoren ein unterstütztes Daten-Format in die MySQL Datenbank

importieren können gewährleisten das ein älterer Stand eines Datensets erhalten bleibt

und nur das neue in die Datenbank hinzugefügt wird, dabei evtl.

einen kurzen Konsistenzcheck durchführen (ein

Überlappungsbereich muss mit dem entpsrechenden bestehenden

DB Inhalt übereinstimmen

Figure 65: Parts of the original technical requirements catalog document (in German).

Visualization of virtual Knowledge Communities 187

8.3.2 Software Architecture

From the lists of requirements the actual architectural software approach needs to

be derived. The resulting software framework for the Social Network Intelligence

Tool is depicted in Figure 66. The main constituents are the frontend including the

Graphical User Interface and the intelligent backend including the connectors to

the original data sources. These two components are introduced in more detail in

the next section. In general the connector accesses the original data source of elec-

tronic expert communication. This can either be a flatfile dump export of an ar-

chive, a database access, or an online HTML, FTP49, or NNTP access. The data is

transformed into a standardized format and stored in an integrated database. The

data analysis and management is carried out by a backend component which for

example starts the textual analysis. This backend also provides a very rich inter-

face to the frontend which allows accessing the authors, their properties, the rela-

tionships, filtering results, or accessing the database etc. The frontend generates a

sociomatrix (matrix) and a graph. Via dynamic filtering and visualization render-

ing the intended graph is computed and displayed on the screen in the selected

visualization form. Via the graphical user interface (GUI) the user can manipulate

the rendering and filtering of the network to configure his individual perspective

on the community network.

Figure 66: General Application Framework with Backend and Frontend. The communica-

tion data is elicited by Connectors and visualized by the Frontend.

This rather abstract architectural level can be broken down into the actual UML

class diagram, which gives more insights about the various classes and how they

interact to conjointly produce the desired graph and analysis. It includes data

classes (like for buffering discourse or author data), data representation classes

(like the complete graph), interface implementation classes (like the individual

connectors or the 3D Renderer), interfaces to define the implementations, control

49 File Transfer Protocol

ELECTRONIC

EXPERT

COMMUNI-

CATION

GUI

Connector

(Data

Extractor)

socket:

jdbc

http

nntp

file

BACKEND: Automated Data Preparation, Management, Analysis

Backend

Data

Analysis

and

Manage-

Ment

Buffered

Network

Class

Structure

Sociomatrix

Generator

and

Layouting

Graph

Dynamic Filtering

and Manipulation

Visualization

Renderer

2D-FR

3D-FR

J3D-FR

2D-Circle/FR

TIMEAUTHOR RELATION

FRONTEND: Automated Data Visualization and Manipulation

TEXT

Abs-

trac-

tion

Layer

TEXT Analysis

Integrated

Data-

base

(MySQL)

188 M.Trier

classes which host the application logic, and user interface elements. Figure 67

summarizes this class structure.

GuiMain

-messegesReceived : Ganzzahl

-messagesSent : Dezimal

-FirstParticipation : Datum

-LastParticipation : Datum

Author

-Name : Zeichenkette

Discourse

-Time : Datum

Message

-MessgesNumber : Ganzzahl

-FirstCorespondence : Datum

-LastCorespondence : Datum

Relation

-X : Ganzzahl

-Y : Ganzzahl

-Z : Ganzzahl

-Size : Ganzzahl

Node

-weight : Ganzzahl

Edge

SocialMatrixVisualization

+drawVisualisation()

+setLayout(GraphLay()

+setGraph()

+setDrawNodeLabels()

+setNodeLabelSize()

+setZoom()

+setMaxNodeSize()

+getRenderComponent()

+getLegendComponent()

+layoutHasChanged()

+setDrawEdgeLabels()

VisualizationRenderer

GraphLayout

FRLayout FrLayout3D

Basic2dRenderer Basic3dRenderer

RenderComponent

Graph

****

TimeVisualisation

LegendComponent

***

Interfaces

Gui Elements

Data Classes

Data Representation

Classes

Control Flow Classes

-Ende11

-Ende2

-Ende31

-Ende4*

Implementation of

Interface

TimeVisualisationTemplate

-Ende5

-Ende6*

CommetrixData

Diskurse Quelle

Connector

NewsgroupConnector InstantMessangerConnenctor HandyConnestor

-Ende7

-Ende8*

-Ende9 *

-Ende10

Properties

-Ende111

-Ende12*

Figure 67: Class Structure of the Software for CoP Visualization and Measurement.

Visualization of virtual Knowledge Communities 189

8.3.3 Data Sources and Data Architecture

In order to determine the data architecture, the structure and elements of the origi-

nal data sources have to be analyzed. This has then to be mapped into a data struc-

ture, which is suitable for Social Network Analysis. As guidance, Scott (1991) de-

scribes three types of data in Social Network Analysis:

Relational Data: This is the representation of contacts, ties and connections be-

tween the members of a social network. These relations link one participant of the

network to another.

Attribute data: Attributes describe the participants in a network. They may be

collected by questionnaires. Attributes might be sex, age or rank in an organiza-

tion.

Ideational data: Describes meanings motives, definitions and typologies of net-

work members.

Looking back at the source data, a very widespread format to store virtual discus-

sions is the Network News Transport Protocol (NNTP) standardized in RFC977

(Kantor and Lapsley, 1986). It is the foundation for internet-based newsgroups

and has been employed in the project as the primary data source for establishing

the data structure. However, later it has been extended to accommodate for more

modern and comprehensive data attributes of other discussion board systems. The

NNTP standard defines only very few elements for storing an expert communica-

tion network on the newsgroup server. They include a unique message identifica-

tion string, the user name, the posting topic, and the posting body. Useful data

items, but not captured are passive readers of a posting or topic keywords.

This NNTP format is now being introduced in order to discuss the typical appear-

ance of a data source of a virtual communication network. Alternatives to this

standard are lose HTML formats, XML50 structures, or databases.

NNTP is no special software application, rather it is a standard which allows ex-

changing articles (i.e. distribution, inquiry, retrieval, posting) using an interactive

internet connection. It is employing a reliable stream-based connection (usually

TCP51) between a client in the network and a server residing on a host, which

stores the postings. Various standardized commands enable the client to receive or

to send articles. NNTP is text-based. Like with SMTP52 for mail exchange, there

are defined headers to coordinate the exchange and contents (or body elements),

which belong to a header. This header structure is the standard, which determines

the available data structure and hence subsequently also the design of the data-

base. Therefore the main elements of NNTP structures will now be outlined.

50 eXtensible Markup Language

51 Transmission Control Protocol

52 Simple Mail Transfer Protocol

190 M.Trier

Articles start with a series of required headers and can have optional headers. The

latter should start with the letter X (e.g. X-Complains-To) for better recognition of

its optional character. Among the required headers for analyzing expert networks

are:

• From: The from header contains the e-mail address of the author. Its cor-

rectness is not verified. Usually the name of the author is also contained

(in round brackets) in this segment, e.g. ‘FROM: author1@mail.com

(Author Name)’.

• Newsgroups: News can be published in the server’s hierarchical list of

groups. This publication target is given after the newsgroups header.

Usually, only one target position is defined, as cross-postings (‘xpost’)

are often not wanted. An example for such a destination is

‘de.comp.lang.php.installation ‘.

• Subject: The subject header contains the title of the message. It should

help a user to identify the contents of that message. If the message is a

reply to another contribution, this fact should be signaled by inserting

“Re:” at the beginning of the title.

• Date: The date header stores the time stamp of sending the message. It

depends on the system clock of the client. An example is ‘Date: Fri, 13

Feb 2004 01:29:35 +0100’.

• Message-ID: The first server which receives the message from the client

assigns an internal unique identification code. Together with the com-

plete name of the server, this results in a unique message-ID. Other serv-

ers can thus reference to this message. They are not allowed to change

this identifier. An example is ‘4p4o20tiva8b1[email protected]m’.

Next to the required headers, there are optional headers possible. The most impor-

tant one of this group is the references header. Discussions in the Usenet are or-

dered in a hierarchy. There are initial requests, which get answered by subsequent

contributions. These answers can get comments and so on. This hierarchical struc-

ture is derived from the references header. It simply contains all messages, to

which the current posting is either the direct or indirect answer. Message-IDs are

stored in the sequence of the discussion. For example: ‘References:

<c0em8k$ae[email protected]e> 4p4o20tiva8b1k6v7[email protected]’.

To elicit the actual people network from the postings, these references between

postings are extracted and analyzed. They indicate answers or comments to a pre-

vious posting and hence a communication relation between two persons. These

relations between authors are the fundamental information for creating an expert

network from the data set. An example for such a hidden communication relation

between two authors of a NNTP-based discussion group is shown in Figure 68.

Visualization of virtual Knowledge Communities 191

Figure 68: Relation between online Discussants in NNTP.

For accessing information about a discourse, a series of related standardized

commands is available. They include:

HELP: Lists all available commands

LIST: Returns a list of all available newsgroups

NEWNEWS: Returns a list of all new article-IDs

NEWGROUPS: Returns a list of all new newsgroups

GROUP: Specifies newsgroup for next commands

ARTICLE: Return an article

BODY: Return the text of a selected article

HEAD: Return the headers of a selected article.

These commands can now be utilized to code a program which extracts the re-

quired messages, their authors, references, time stamps, contents etc. of a selected

newsgroup from an available online news server. The resulting data can then be

stored in a database. Figure 69 shows an example of how the extracted data looks

like. The message-ID, author name, author e-mail, posting subject, bodytext, post-

ing time, and referenced message-ID have been extracted.

549

Anywhere

toff@estang.com

Thu, 10 Jul 2003 10:57:10 +0800

Apache2.0 can't start?

...I cant'st start Apache2.0 service...

<3f0cd63e@newsgroups.borland.com>

550

Hans Karlson

hk@hkarl.com

Thu, 10 Jul 2003 13:40:43 +020

Re: Apache2.0 can't start?

...put thefilenamebetweenquotes...

<3f0d50bb$1@newsgroups.borland.com>

<3f0cd63e@newsgroups.borland.com>

Post-Nr.

Username

User-Adress

Post-Date

Post Topic

Post Content

Post ID

Post-Nr.

Username

User-Adress

Post-Date

Post Topic

Post Content

Post ID

Refers to Post ID

192 M.Trier

Figure 69: A Snapshot of Information extracted with a first Prototype

of a Newsgroup Connector.

This table is not capable to satisfy all requirements (i.e. speed of massive queries)

of the final tool. Therefore a more sophisticated data structure has been designed

and implemented as a MySQL database. The main data elements are the discourse

itself, the messages, and the authors. Added entities are keywords, which can be

stored as additional information about messages, further subjects and a table for

dynamic properties. This table illustrates the idea, that not all data to be extracted

can be a-priori projected by the database designer. This renders it necessary to in-

clude a generic element, which stores discourse specific information about mes-

sages or authors. The database realizes this issue by offering a generic meta-data

table for messages and authors (cf. MessageMetaDataTypes and AuthorMeta-

DataTypes in Figure 70). The designer of a specific connector for an individual

discourse can create new properties and subsequently store the extracted data in

the database. For this, the connector first creates a new meta-data type in a special

table and then adds entries into the metadata table which reference this type fol-

lowed by the actual instance of the data element. For example, if user location

shall be included into the database, the connector would first need to instantiate

the new type Location and then it simply adds new metadata entries, which con-

tain the specific location (e.g. US, Poland, Germany, etc.). The following Figure

70 shows the final Entity-Relationship-Diagram (ERD) of the database design.

Visualization of virtual Knowledge Communities 193

Figure 70: Entity-Relationship-Diagram of the Data Structure.

The final database architecture is shown in Figure 71. If a message has more than

one author, the table MessageRecipients stores these relationships.

Figure 71: Final Database Architecture of the Tool.

Keywords

MessageMeta

DataTypes

Messages Authors

Discourses

Body

AuthorMeta

DataTypes

Message

Has

Value

Subject

Author

Has

Value

Belongs

To2

Belongs

To1

Active

Written

0..1

Alias

194 M.Trier

8.3.4 Intelligent Backend and Connectors

After establishing the data structure for storing electronic discourse data sets, the

actual software components which work with this data can be introduced (also cf.

Figure 66). The first major cluster of classes constitutes the intelligent backend

component. Its main elements are modules for source import via connectors, filter-

ing, data base handling, data buffering, and the interface for the frontend’s re-

quests. One of its major tasks is to execute available connectors which can be se-

lected by the user. The approach of using a set of connectors recognizes the re-

quirement to enable the access to a wide variety of information sources. For this, a

modular approach has been adopted, which allows for easily creating individual

connectors for every new data source. Their objective is to elicit useful data about

the information exchange between participants of some type of online discourse.

Currently, the solution provides implementations of connectors for importing e-

mail archives, instant messaging protocols, newsgroup archives, and Slashdot dis-

cussions. However, the idea is that new connectors can simply be added when

needed. In order to minimize the effort for coding such new connectors a ‘data

base abstraction layer’ has been introduced. The connector can work directly with

an interface of Java objects and does not have to take care about details of the da-

tabase. As introduced in the previous section, the database provides for generic

(‘dynamic’) properties which can be created and filled according to the individual

needs of the connector writer. This allows storing virtually any extractable data

which might be of interest. Later, the frontend automatically identifies all proper-

ties and integrates them into selectable choices for color-coding, node-size-

indicator, edge label, and node label. By this, the software offers individual analy-

sis based on one integrated database structure and without any modifications of

the Graphical User Interface. For example, in Slashdot archives, the user can se-

lect the author evaluation, which does not appear when a newsgroup discourse is

analyzed.

Visualization of virtual Knowledge Communities 195

Figure 72: Integrating different Storing Paradigms in electronic Discourses.

In the process of connecting and importing data sources, another fundamental de-

sign issue is of importance. In electronic communication networks, there are two

basic storage paradigms: peer oriented and hierarchical networks. Whereas instant

messaging, e-mail, or chats belong to the first category, discussion groups include

hierarchical information, as messages can refer to parent messages. Obviously the

hierarchical paradigm comes along with the central mode of online communica-

tion, whereas decentralized online communication is stored in peer-to-peer para-

digm (also compare section 6.3). In Figure 72, the different storage paradigms are

visualized. In peer oriented archives like those of e-mail servers, author A is con-

tacting author B directly via the message. However, in hierarchical archives like in

discussion groups, author A is writing message 1 which references message 2,

which has been written by author B. Basically, the latter procedure is comparable

with author A sending a message to author B, however the references are stored

differently. The hierarchical storage paradigm maintains the information which

message is an answer and which is an initial request. This allows to virtually trac-

ing back a thread or dialogue. Such referential information can not be automati-

cally extracted from peer-to-peer storage, which therefore becomes more frag-

mented.

The connector recognizes these two different storage paradigms and transforms

the data in a way that it can be stored in one single integrated data structure intro-

duced above. An example is given in Figure 73. This procedure leaves room for a

very important potential. By transforming multiple storage approaches into one

structure, the tool is theoretically capable of integrating different channels of

communication into one analysis. This concept meets the challenge of an inte-

grated analysis of multi-channeled Communities of Practice, introduced in chapter

6.3.

Peer-to-Peer

Messages are sent directly

between authors

Joe writes Message to Jack

The message establishes the

relation between Joe and Jack

Threaded

Messages are answers to other

messages

Brian writes a Message

Michael writes a Reply

The fact that Brian writes a

reply establishes the relation.

Author Authorcontacts Message Parent Msg.references

Author Author

Message

196 M.Trier

Figure 73: The Paradigm Transformation Approach.

Another element of the backend is comprised of a set of filtering classes. It is in-

cluding filters for authors, keywords, relationships, and time-periods. They are ac-

cessible via a sophisticated backend interface. Filtering the datasets is often neces-

sary to reduce the amount of elements in the visualized graph, because large

communication networks can easily include many hundred authors and even more

relationships. No transparent graph can be drawn for such a sample. This is why

dynamic and static filters are included.

Among them, author filters allow for limiting the number of authors in the dataset.

This can be utilized to explore the various ego-networks in a social network, i.e. to

observe the direct environment of a single selected author. Further, filtering for

authors can show, if two authors are directly connected.

Relationship filters set a threshold for the strength of the relationship, i.e. the

number of messages between two persons. Relationships which are less intense

than predefined by the user are not shown in the final graph. Thus, the network

only contains edges for strong relationships and weak connections are invisible.

This reduces the complexity of the overall network.

Time-period filters reduce the visible network by limiting the start and end of

communication activity. This filter is also used as a moving window and thus sup-

ports longitudinal analysis of the network’s evolution, which is described in more

detail in section 8.4.2.

Finally, keyword filters can reduce the network size by only showing a subnet of

the part of the communication, which is including a certain keyword or topic. This

feature is closer examined in chapter 8.4.4.

Peer-to-Peer Hierarchically threaded

JackJohnMsg1

RecipientAuthorMessage

JackJohnMsg1

RecipientAuthorMessage

Msg2BrianMsg3

MichaelMsg2

ReplyToAuthorMessage

Msg2BrianMsg3

MichaelMsg2

ReplyToAuthorMessage

JackJohnMsg1

Recipient

Msg2MichaelMsg3

BrianMsg2

ReplyToAuthorMessage

JackJohnMsg1

Recipient

Msg2MichaelMsg3

BrianMsg2

ReplyToAuthorMessage

Visualization of virtual Knowledge Communities 197

The following Figure 74 gives a partial overview about the according backend in-

terface responsible for time-period filtering. The processed requests of the fron-

tend can include the definition of start and end dates of filtered discussions or the

first and last date of stored data. Further commands of this interface are utilized to

add messages at the end or at the beginning of the selected period, if they are in

the filtered set or to remove them if they are not.

+setFilterMaxDate(Zoll Date) : FilterResult

+setFilterMinDate(Zoll Date) : FilterResult

+getFilterMaxDate()

+getFilterMinDate()

+addLastMessages(Zoll int) : FilterResult

+removeLastMessage(Zoll int) : FilterResult

+addFirstMessages(Zoll int) : FilterResult

+removeFirstMessages(Zoll int) : FilterResult

Discourse

+getNextAuthor()

+getNextRelation()

FilterResult

Figure 74: Time-period filtering Interface and Functions.

The architectural separation into a backend and a frontend component requires a

comprehensive interface between the backend and the frontend. The different fil-

tering requests discussed above already provided a first example of it. Generally,

user manipulations (as GUI events) and requests are submitted to this interface

and the backend returns the according information or executes the according

commands. This modularization is useful for a later extension with a web-based

client, which can then communicate with the original backend. Figure 75 now

gives an overview about the remaining methods of the backend’s interface to the

frontend. For example, the user wants to know the number of relations of an au-

thor and the average strength. The frontend is accessing the interface to activate

functions for counting the relations of the author or the number of messages in the

relations. The backend returns the necessary information and the frontend shows

them to the user. In order to improve the application’s performance for database-

related requests of the frontend, the database’s data is buffered in buffer classes in

the backend. This reduces the overall number of database accesses and connec-

tions, which results in much faster response and processing times.

198 M.Trier

Commetrixdata

+getAllDiscourses()

+bufferData()

+getAmountOfAuthors()

+getAmountOfMessages()

+getAuthorList()

+getMessageList()

+getRelationList()

+getRelations(Zoll Author)

+getRelation(Zoll Author, Zoll Author)

Discourse

+countRelations()

+getDiscourse()

+getFirstParticipation()

+getLastParticipation()

+getInDegree()

+getOutDegree()

+getMessagesReceived()

+getMessagesSent()

+getName()

+getRelations()

+getRelationToAuthor()

+getRepliesReceived()

+isActive()

Author

+getAmountOfMessages()

+getAmountOfMessagesSentBy(Zoll Author)

+getAuthors()

+getFirstAuthor()

+getSecondAuthor()

+getMessages()

+getMessagesWrittenBy(Zoll Author)

+getMessagesWrittenByFirstAuthor()

+getMessagesWrittenBySecondAuthor()

+isSelfRelation()

+isActive()

Relation

+getDate()

+getDiscourse()

+getAuthor()

+getParentMessage()

+getRecipient()

+getRelations()

+isActive()

Message

Figure 75: The Backend’s Interface to the Frontend.

8.3.5 Frontend and User Interface

The frontend is responsible for generating visualizations using the data prepared

and provided by the backend. Further, it presents these visualizations in an ergo-

nomic Graphical User Interface (GUI). The user can manipulate the configuration

of the visualization to represent individual information or reduce the visible size of

the network. For these functions, the main necessary elements of the frontend are

the Graph, the Layout, the Control, the Renderer, and the GUI component.

The Graph element is storing the data set for the visualization with the core ele-

ments node and edge. Much more complex is the Layout component. It is apply-

ing the Spring Embedder Algorithm introduced by Fruchterman and Reingold

(1991) in order to create a meaningful graph from the information about the com-

munication between the experts. The method is applied to present a (people) net-

work as a structure with nodes and edges. Nodes represent actors and edges repre-

sent the relationships between them. Actors that are shown close together have

stronger links than actors which are very distant. The abstract layout component

can get instantiated to allow for individual layouts to present the information. This

enables a later addition of completely different views. Currently, layouters are ap-

plied to generate two dimensional and three dimensional representations of the

communication structure. The following Figure 76 shows an example. The illus-

trated network structure is a preliminary result of a 3D layout. It shows the authors

as nodes and their relationships as edges. The details of the underlying algorithm

are introduced in the next chapter about the applied technical frameworks meth-

ods.

Visualization of virtual Knowledge Communities 199

Figure 76: Spatial Layout of Nodes constituting a preliminary communigraph.

The data and flow control element for visualization management is the compo-

nent, which actually activates the necessary computational tasks. It is thus a coor-

dinating and steering class intervening between the other ‘specialist’ classes. The

general process executed by it starts with the creation of a graph (class) according

to the selected data. Then, it lets the layouter compute the spatial coordinates. Fi-

nally, the renderer is activated to present the nodes and edges on the screen. Every

step does not necessarily need to be completed before the next can start. This al-

lows for changing the data in the graph or its layout while simultaneously the user

can follow the according changes. This results in animated movements of nodes

which move to their new or final positions. The control element can further ma-

nipulate the graph according to the filters set by the user. They include a time filter

to show only the nodes which are active in a given time period. This allows for

longitudinal analysis and animation of the evolvement of the networks (also see

chapter 8.4.2).

Last but not least, the application includes a set of renderers, which can be se-

lected by the user to switch from 2D to 3D. However, not every renderer requires

its own layouter. For example, to deal with complex networks which use up large

amounts of memory and processing time, there are two modes of 3D visualization:

a Java3D based rendering machine with light sourcing and a ‘simple’ or ‘emu-

200 M.Trier

lated’ 3D visualization which in turn saves considerable amounts of memory and

processing time and can be employed for larger networks, where the full 3D vari-

ant would be too slow. Both utilize the same 3D renderer to calculate the accord-

ing coordinates.

The last important component is the graphical user interface. It is necessary to

manipulate and configure the generated visualizations in order to create individual

views which help the user to understand his network under observation. As most

insights will not be generated by the initial graph layout, but by filtering, color-

coding, labeling, or other means of configuration, a sophisticated interface is re-

quired which creates the impression that the user can actually play with his model

to learn about the underlying structure. Figure 77 summarizes the GUI elements,

which help the user to control the application. It also gives a first impression of the

available functions of this book’s software approach.

Figure 77: User Options for controlling the Application.

With these functions, the user starts with importing a new data source or selecting

it from the database. He can select his preferred visualization form (e.g. 2D or

3D). Then he can change the layout by zooming or changing optimal distances or

movement behaviors of nodes. The user can further configure the view by select-

ing the label of the nodes (e.g. index number, name, e-mail, or keywords), the

property which is translated into a color code (e.g. blue for poor evaluation or or-

ange for good evaluation), the property which is translated into circles surrounding

the node (e.g. hierarchy levels, only in 2D), or the edge label (e.g. number of ex-

changed messages). To build subsets he can utilize time-based or ego-based fil-

ters. The latter allow filtering out a subset of authors. Further, the animation of the

Layout

Zoom

Group Distance

Edge Length

Max Node

Movement

Navigation

(e.g. Rotate)

Configuration

Select

Node Label

Select Node

Label Size

Select Factor for

Node Size

Select Factor for

Ring Represent.

Select Factor for

Color Represent.

Select

Edge Label

Select Factor for

Edge Length

Subsets

Select Time-based

Visualization

Animation Speed

Node Movement

Behaviour

Max Node

Movement

Select/Filter

Author Subsets

Start Animation

Select/Filter

Keyword Subsets

Stats

Discourse

Information

Author

Information

File

Import new

Data Source

Select Discourse

Delete Discourse

From Database

Export

Visualization

FR 2D Layout

Simple 3D Layout

Java 3D Layout

Monitor

Relationship

Information

Measurement

Report

View Tables

Author Tables

Relationship

Tables

Message Table

Help

Guide

About

Visualization of virtual Knowledge Communities 201

network evolution can be controlled. Finally, the user can access the measure-

ments and analysis panel (‘stats’) which contains important information about the

discourse, the author, the relationship and the measurement report according to the

measurement system developed in chapter 7.2. In the table menu, the analyst can

actually observe the original tabular data. This can be used to add more properties

manually, like for example the department, the author belongs to, or a project.

Such information could subsequently be used to color the nodes according to such

affiliations.

Figure 78: Inserting Properties from the Backend using a dynamical combobox Menu.

The dynamical aspect of the application can be illustrated with the handling of

properties. Every individual application that hosts discussions stores different ad-

ditional data, for example message evaluations, author evaluations, author pro-

files, or author login times. As introduced in chapter 8.3.4, the intelligent backend

offers a means for storing such discourse, message, or author properties regardless

of the individual data source. Now, the frontend automatically identifies such

stored properties of the active discourse and offers various means for manipulating

the produced graph. Nodes as well as edges can be labeled, sized, and colored to

represent all meaningful information. It automatically detects, that e.g. the prop-

erty ‘author is member of core group’ can only have the two states yes or now, or

that the evaluation for a certain discourse reaches from -1 to +5. The mapping is

dynamically being aligned with this information. This is done by retrieving certain

property types: some are enumerations of strings which do not imply a scale, some

convey ordinal scales, and some are Boolean, i.e. true or false. The frontend auto-

matically enters qualifying properties into the according combo-box for the user to

select it. This feature is shown in Figure 78. Using this dynamic property map-

ping, every individual property of a discourse can hence be maintained, analyzed

and visualized. The complete user interface is illustrated in Figure 79.

202 M.Trier

Figure 79: A Sample of a final User Dialog as presented by the Frontend.

8.4 Applied Technical Frameworks and Algorithms

The software components illustrated in the previous sections require underlying

methods, technical approaches, algorithms, or available libraries. Key methods are

graph theory and the related spatial layout algorithms to incorporate social net-

work visualization and analysis. Key technical elements are 3D computer graphics

as well as document and media export. To extend the core visualizations, further

domains have been incorporated. Longitudinal network visualization offers the

observation of time-related evolvement of the network clusters. Further, content

coding is a technique to create more insights about the content of the discussions

in the network. It has to be differentiated from automated keyword extraction as

another approach to generate such insights: Whereas content coding gives insights

about the intended purpose of communication (help, flaming, request, social, etc.),

keywords are actually representing snippets of the contents, which can be helpful

for searching the net or creating meaningful subsets. Finally, an initial part of the

complex measurement system as introduced in chapter 7.2 has been developed

towards a technical framework, which subsequently has been added to the soft-

ware application.

8.4.1 Graph Theory and Information Visualization

A fundamental requirement of this work’s approach is the issue of how to display

a network of nodes and relations and how to include or visualize measurements in

Visualization of virtual Knowledge Communities 203

meaningful graphs. The required process to generate such visualizations from the

data sources has been theoretically founded on the process proposed by Card et al.

(1999:17). In their model, raw data is transferred into data tables which are a more

accessible structure for visualization. Data tables must be mapped into a certain

visual structure. This visual structure should make the data as easy and as efficient

to interpret for the viewer as possible. The structure is later observed by the user

via different views (cf. Figure 80). They allow understanding a complex system

(of information) by watching it from different perspectives. All these steps can be

augmented by human machine interaction to make the visualization as specific

and as insightful as possible.

Raw

Data

Tables Visual

Structures

Views

Data

Transformations Visual

Mappings View

Transformation User

Human Interaction

Figure 80: Diagram of Reference Model for creation of a Visualization.

Source: Card et al. (1999:17)

The graph which is to be generated via such a process can be theoretically derived

from and described with the terminology of graph theory. According to Battista

(1999:3), a Graph G = (N, E) consists of a finite set N of nodes and a finite set of

edges that are constituted by pairs (u, v) of nodes. An edge with u = v is called a

self loop. Edges that occur more than once are called multi edges. A Graph that

has no self loops and multi-edges is called a simple graph. If there is a graph G' =

(V', E') where V' ∈ V and E' ∈ E, then G' is a subgraph of G. A drawing of a

graph is planar if no two edges cross each other. A graph G is planar if it permits a

planar drawing. Planar graphs are important for displaying graphs because cross-

ing edges decrease readability of graphs.

The last point directs the attention to the importance of ergonomic factors in graph

drawing. Here, the theory of Gestalt Laws comprises a set of guidelines which

can aid in the generation of insightful and user ergonomic graphs. These Gestalt

Laws examine several rules that describe the way humans see patterns (if not

quoted otherwise, mainly taken from Ware, 2000). First, things that are located

close together will be grouped together as a perceptional unit. Second, elements

with the same shape are grouped together. Slocum (1983) introduced the principle

of spatial concentration in which people perceptually group regions of similar

element density. A further aspect for the design of visualizations is that humans

are more likely to see a smooth continuation as a better continuation of a form

than an abrupt angular one. Palmer and Rock (1994) argue that connectedness of

204 M.Trier

structures results in a stronger perception of relation to each other than size, color

and shape. Symmetry of graphs plays an important role, as symmetrically ar-

ranged visual structures are much easier to recognize as a whole. Closed contours

will be seen as one object. When seeing a closed contour, the viewer tends to di-

vide the observed space into inside and outside of the contour. Smaller compo-

nents in a pattern are recognized as objects. A figure is something object-like, per-

ceived to be located in the foreground. Everything else is identified as background

(ground) behind this figure.

A further issue related to Gestalt laws and user ergonomics of visualizations are

drawing conventions, which define certain requirements for a graph to fulfill in

order to belong to a specific type of representation (Battista, 1999:12). These

graph drawing conventions include:

• polyline drawing, where edges are drawn in polygonal chain of lines,

• straight line drawing, where edges are drawn as straight lines,

• orthogonal drawing, where edges are drawn as chains of alternating con-

nected horizontal and vertical lines,

• grid drawing, where nodes and edge bindings have integer values, and

• planar drawing, where there are no edges crossing.

For the graphical visualization of people networks, these graph drawing conven-

tions can be examined to create one of these pre-defined classes of graphs. Exam-

ples for the resulting visual appearance are given in the following Figure 81.

a) b)

Figure 81: Example Graph Drawing Conventions for specified Types of a Graph:

a) polyline b) straight line c) orthogonal.

Source: Battista (1999:13)

Next to such types, aesthetical properties can support the generation of a readable

graph. Among the basic applicable aesthetical criteria for graph drawing are (Bat-

tista, 1999:13):

Visualization of virtual Knowledge Communities 205

• Crossings: Minimization of crossings in a graph. Ideally a graph drawing

should be planar but this is not always possible.

• Area: Minimizing the area of drawing. This is only of meaning if the

graph drawing cannot be scaled down.

• Uniform edge length: Minimization of the variance between edge

lengths.

• Angular resolution: Maximization of the smallest angle between two

edges incident to the same node.

• Symmetry: emphasize and display symmetry of a graph in its drawing.

For drawing graphs, it must be considered that aesthetical criteria are often con-

flicting each other. Thus tradeoffs are unavoidable. Further, it can be difficult to

deal with all chosen aesthetics at once even if they don't conflict (Battista,

1999:17). Graph drawing algorithms therefore generate priorities between the ap-

plicable aesthetical rules.

Next to these basic rules or requirements for designing insightful graphs, visual

variables which convey information have been researched and systematized by

Bertin (1983:42). He identifies the following six items, which can augment a

graph like the communigraph of this book’s software approach in order to repre-

sent additional information:

• texture of an object,

• size of an object,

• color of an object.

• orientation of a pattern or an object from vertical to horizontal,

• color value of an object, which is the continued variation of color from

black to white. It may also include shades of grey or of any other color,

• shape of an object, and

• plane or position of an object in a two dimensional planar space.

Summarizing, the theoretical process of graph drawing can aid in the generation of

visualizations from raw data. Resulting graphs can be described theoretically us-

ing the terminology of graph theory. This allows for a systematical analysis of

graphs in a subsequent step, as measures are founded on the theoretical structure

of a graph. There are many further concepts which generally guide the design of a

computational graph. Among them, there are Gestalt Laws, types of graphs, aes-

thetical criteria, and options for encoding additional information. These abstract

theoretical concepts are now involved in the subsequent generation of an insight-

ful graphical representation of data about people (or social) networks. After de-

scribing the specialization of graph theory in the field of social network visualiza-

206 M.Trier

tion and analysis, the next chapters introduce the concrete graphical layout algo-

rithm adopted to recognize the discussed theoretical requirements for an insightful

graph, discussed in this chapter and the application of a technical environment

which helps to actually draw and represent the resulting network graph on a com-

puter.

8.4.1.1 Social Network Visualization and Analysis

Based on the concept of sociomatrices for network analysis as introduced in chap-

ter 7.1.3, analytical approaches concentrating on social network graphs have been

developed. They enable a visual analysis of large people networks. This social

network visualization is a specialization of the more abstract graph theory. Scott

(1991:70) describes some basics of graph theory in the domain of social networks

by using very simple undirected binary graphs consisting of nodes and edges be-

tween them. In social network visualization, the nodes represent the authors (per-

sons) and edges represent their relations. The term ‘undirected’ refers to the miss-

ing information about the direction of the edge, which does not point to a particu-

lar node but connects them. ‘Binary’ refers to the possible value of the edge: an

edge either exists (1) or not (0). If two nodes are connected by an edge they are

called ‘adjacent’ to one another. All nodes adjacent to a node comprise its

neighborhood. The number of nodes in such a neighborhood is called the node’s

‘degree’. In other words, a node’s degree is a measure for the size of its neighbor-

hood. In analogy to the sociometrical approaches described in 7.1.3, in directed

graphs there is a difference between edges pointing towards the node (‘indegree’)

and away from the node (‘outdegree’). Nodes can be directly connected by an

edge or indirectly by a sequence of edges. Such a sequence is also referred to as a

‘path’. A path’s length is measured by the number of edges in it. The length be-

tween two nodes is the length of the shortest path that connects them.

Figure 82 illustrates how the concrete graph can actually be extracted from an ex-

isting discussion group. The procedure described is simultaneously following the

process from raw data to views suggested by Card et al. (1999:17) and depicted in

Figure 80. As chapter 8.3.3 described, discussion groups follow a hierarchical

paradigm to store their communication. This means author 1 writes message A,

which is referenced by the answering message B, written by author 2. Thus, a rela-

tion between authors 1 and 2 is implicitly existent if there is a relation between

messages A and B (also compare the upper-right illustration of two newsgroup

headers which are connected and thus also connect their authors in Figure 82).

On the left hand side of Figure 82, the hierarchal storage pattern of a discussion

board is visualized. User 1 writes a message and user 2 answers it. This is also in-

dicated by his subordinate position in the tree’s hierarchy. User 3 comments the

answer and thus provides an answer to user 2. Then user 1 intervenes again and

answers on the comment of user 3. User 4 posts another answer to user 2 (same

hierarchical rank as user 3 but subordinate to user 2), user 5 posts his answer to

Visualization of virtual Knowledge Communities 207

the request of user 1, just as user 3 does. In such a structure, it can be differenti-

ated between various types of relationships. Another person can either be directly

referred by another person or directly reference that person’s contribution. Further

he can indirectly reference some person (like user 3 who is indirectly also posting

a contribution to the initial request of user 1) or be referenced.

Figure 82: From Message Hierarchies to Relationship Networks.

The actual technical reference between communication acts is depicted in the up-

per right corner. Every posting is equipped with a header, which contains the ref-

erence to the related posting (i.e. posting 3f0d50bb$1 refers to posting 3f0cd63e).

If the relationships for person 1 as implied in the figure on the left hand side are

translated into a network graph consisting of nodes and edges, the two graphs in

the bottom-right corner of Figure 82 appear. The left network graph contains di-

rected edges, where different edges mark indirect direct references. The design

question is how to generate the overall relationship between two authors from the

multitude of different message relationships. To increase transparency and reduce

visualization complexity in the resulting computer generated graphs, it is advis-

able to create undirected relationships. This can be done by summarizing the set of

directed relationships or by only counting sets of two subsequent messages in an

opposite direction (or in other words a dialog). Indirect references can either be

ignored or receive a small weight, which affects the strength of the overall rela-

tionship. The second network on the bottom-right corner is showing the weighted

undirected relationships of person 1 if indirect references are ignored. A strong

relationship to user 3 can be observed. It can also be verified in the discussion tree

549

Bob

[email protected]

Thu, 10 Jul 2003 10:57:10 +0800

Apache2.0 can't start?

...I cant'st start Apache2.0 service...

<3f0cd63e@newsgroups.borland.com>

550

Hans-Dieter X

hdx@hdx.com

Thu, 10 Jul 2003 13:40:43 +020

Re: Apache2.0 can't start?

...put the filename between quotes...

<3f0d50bb$1@newsgroups.borland.com>

<3f0cd63e@newsgroups.borland.com>

Post-Nr.

Username

User-Adress

Post-Date

Post Topic

Post Content

Post ID

Post-Nr.

Username

User-Adress

Post-Date

Post Topic

Post Content

Post ID

Refers to Post ID

User 1

User 2

User 1

User 3

User 4

User 5

User 6

User 3

User 6

User 3

User 1

User 6

User 5 User 4

User 3

User 2 User 1

User 6

User 5 User 4

User 3

User 2

weighted strength,

undirected relationships

208 M.Trier

on the left: user 3 is answering one request, user 1 is giving feedback to user 3,

and later user 1 is answering 1 request of user 3.

Based on the resulting network graph, which is simultaneously a social network

visualization as specified by Scott (1991:70), all the available network measures

developed by the research domain of Social Network Analysis are applicable. For

example, the relationship strength is the value of the edge between two nodes in

the graph. If user ergonomics shall be improved, this value can for instance be

aligned with the thickness or the color of the line representing the edge. Reciproc-

ity can be calculated by looking at the different in- and outdegrees of a certain

node in a network. A final example is the network role of an author, which refers

to a visual pattern in the network. For example, in a graph a broker should appear

as a node between two clusters of nodes.

Following these approaches, the final model for this book’s software visualization

approach includes the main elements author and relation (as a quantity of mes-

sages). Authors are represented by a node (sphere). Each node has core attributes

like name, e-mail, and index number. Moreover, the author (node) has several

properties. These are different from core attributes, as they are conveying a quali-

tative or quantitative meaning, which can be grouped, compared or measured.

These properties later allow for comparative analysis of the network’s partici-

pants. Visually, this is supported by enabling the user to encode properties via

node or edge label, node size, or node color.

Relations are an additive set of individual messages and are represented by edges.

Each edge has an index number. Apart from that, it has measurable or comparable

properties, like the number of messages, it represents, the average evaluation of its

contents, or the main keywords.

8.4.1.2 Spring Embedder Algorithm

A very important visualization approach for representing social networks as

graphs is the clustered 2D network graph using the Spring Embedder Algorithm as

proposed by Fruchterman and Reingold (1991). It distributes nodes in a two-

dimensional plane with meaningful separation, which is representing the (distance

of the) original relationship between the nodes. This method has been selected be-

cause it provides a detailed insight into the communication structure of a virtual

people network (e.g. in a discussion group) and suffices many aesthetical criteria.

For example, the original Fruchterman Reingold Layout (FR-Layout) draws edges

at even lengths, keeps the angles between vertices as big as possible, and mini-

mizes edge crossings. Therefore the results have a very clear structure and are

easy to read. The algorithm is very simple and therefore easy to modify. If the

edge length is intended to represent meaningful information (like the relationship

Visualization of virtual Knowledge Communities 209

strength), then the aesthetical element of even edge lengths can be simply omitted

in order to improve the overall utility of the resulting graph.

The underlying mechanism simulates a force system of virtual springs, attached

between authors. Conceptually such force directed methods consist of two parts

(Battista, 1999:304). First, a force system needs to define the physical model

which determines the behavior of nodes and edges. The second element is an algo-

rithm that finds the equilibrium state, which in turn defines when the graph is

ready to be drawn.

To generate the first element - the force system - a matrix is computed, containing

the optimal distance between any two members. This distance is derived from the

strength of their connection. Authors who have a strong relationship are bound by

a higher attractive force and hence should have a smaller distance than authors

with a weak relationship. In the force system, this effect is created by modeling

each node as a charged particle. This means, that repulsive forces are existing be-

tween all nodes to keep them at a distance and simultaneously edges between

nodes invoke attractive forces, so that edges attract two nodes which they connect.

The edge attraction is modeled by the formula

fa d d 2k

and the node repulsion by the formula

fr d k 2d

where d is the distance between the two nodes connected by the edge and k is op-

timal node distance.

Attracting and repulsive forces sum up to a zero at a distance of k. Here, the con-

nection is in equilibrium. In other words, two nodes connected by an edge would

have this distance k if they are in a state of equilibrium and without outer influ-

ence. The developed layout component of the software application is able to let

the user alter k (with the parameter ‘optimal edge length’). Visually, this results in

a scaling effect. The network is laid out on a larger space.

The actual layout algorithm starts by randomly allocating nodes representing com-

munity members on a two-dimensional plane. This results in an initial state with a

random actual distance and the according forces between nodes. Then, the com-

plex system of springs is relaxed. The simulation compares the current with the

optimal distance k. This actual distance can also be expressed as a force: attractive

and repulsive forces are added to determine the current overall forces between

each two nodes (as described above, they should add up to zero if the optimal dis-

tance k is reached). The computed values are stored in a force matrix. Each cell is

indicating the attractive forces fa(d) that reduce a positive difference (i.e. where

the actual difference is still higher than the optimal) or repulsive forces fr(d) that

increase a difference between each pair of nodes in the graph. By adding one

node’s directed forces towards or away from all other nodes, a final force vector

210 M.Trier

can be calculated to move every node for a certain distance into the resulting di-

rection. The amount of movement of the nodes is limited by an upper limit, which

is called temperature. This temperature sinks during the iterations. At the begin-

ning, the temperature is very high because the nodes will be far from their equilib-

rium. But while the layout proceeds the graph gets closer to a state of equilibrium

and the steps of adjustment get finer and finer.

This process is repeated until the complete force system approaches an energy

minimum. This implies that the sum of the differences between the actual and the

optimal distances has been minimized with the current configuration of nodes and

hence the spring system is in its most relaxed position. Visually, during multiple

iterations, a clustered network graph is emerging, showing areas of strong rela-

tionships versus areas where there are no relationships.

The framework for the visualization tool has been designed modular which allows

for subsequent addition of alternative algorithms. In order to understand the im-

plicit characteristics of a graph, it might be very useful to switch between different

algorithms which convey different aspects about the same graph. However, in this

book’s approach, for the beginning, the Fruchterman Reingold layout suffices all

requirements. However, the FR-layout has been extended into a three dimensional

FR-Layout (see for example Figure 76 for a basic layouting result). This is similar

to the 2D approach but includes a third dimension and hence three dimensional

distances and force vectors between nodes. This additional dimension provides

one more degree of freedom for the nodes to find optimal positions and to mini-

mize the overall tension (energy) in the network. This is caused by the fact, that

there is more space on a sphere around a node than on a circle around a node for

his adjacent actors to position themselves optimally. Further, there are less cross-

ings between edges as there is more space for them to position and the three di-

mensional angle between edges in complex graphs are larger, making the graph

more readable and ergonomic. Further, the resulting 3D graph allows for a better

perception and comprehension of the original communication structure, which

now more resembles an object, which can be zoomed or rotated to observe it from

different perspectives, just like a complex molecule. However, there is also a

downside. As there are multiple perspectives possible, it is difficult to have one

optimal view on the graph, as almost always, some nodes are hidden behind oth-

ers.

8.4.1.3 The application of Java 3D

After describing the underlying theoretical methods and algorithms, the derived

technical implementation needs to be introduced. In the software application, two

different technical approaches to present 3D models have been incorporated.

There is a version called ‘simple 3D’ which is simulating the third dimension by

mapping three dimensional edges on a two dimensional pane and by simply in-

creasing the node size of nodes which appear closer to the user. This mode is

Visualization of virtual Knowledge Communities 211

available, when the network is very large and memory resources have to be saved.

If the computer is sufficiently equipped or the network has a reasonable size (de-

pending on the state of the art in memory technology and graphical hardware), a

full 3D mode can be utilized. It is based on the special Application Programmer's

Interface (API) for Java 3D (J3D), an optional addition to the standard Java im-

plementation. Java3D provides 3D rendering for Java programs, but can simulta-

neously use OpenGL as the interface to the hardware. However, J3D does not re-

quire direct hardware device driver support. Conceptually, it is a hierarchy of Java

classes for three-dimensional graphics rendering and includes many high-level

constructs for creating and manipulating 3D geometric objects. All objects reside

in a virtual universe. The classes offer a 3D canvas, which is an extension of a

usual Java Abstract Windowing Toolkit canvas. Instances of special J3D objects

can be created and placed into a ‘scenegraph’. Such a scenegraph is a tree-shaped

hierarchy that specifies all contents of a virtual universe and how it has to be ren-

dered. It contains objects to define issues like geometry, lights, location, orienta-

tion, and visual appearance of objects. Each path in the scenegraph specifies all

necessary information about the bottom element (‘leaf’ without a further child

class) including location, transformation, or behavior. A sample path includes the

creation of a Canvas3D object. A VirtualUniverse Object is attached to it. As its

child class, a Locale object defines the original landmark, where further objects

and properties are located. This Locale object links to a view branch, which con-

tains all necessary information about the view on the rendered structure, including

viewing controls and a ‘movable’ ViewPlatform on which the view ‘sits’ to ob-

serve the content. Further the Locale employs a content branch which contains the

visual objects and their appearance. These two branches are rooted in a special

Branchgroup object, which serves as the root object for scene graphs. Another

special object is the TransformGroup, which is connected to a Transform3D ob-

ject. It can be used to generate transformations like translation or rotation of ob-

jects (relative to the view of the observer). A final option is creating an Interpola-

tor which controls behaviors of objects, like dynamic spinning (which is different

from a static rotation, as the latter is only orienting the object in the universe with-

out a movement).

Based on this conceptual architecture the scene graph in Figure 83 has been speci-

fied for the J3D renderer of the visualization tool for people networks. It starts

with the instantiation of the VirtualUniverse object. The locale comprises the root

and has a simple view branch, which owns a transformation object that zooms the

view according to the user’s commands. The content branch contains two general

transformation nodes. The first is managing the interactive rotation and translation

as requested by the user and the second manages centralization of the structure.

Below these scene graph elements, the actual visual objects are defined, namely

Node3D and Edge3D. The nodes own transformation elements in order to trans-

late the nodes to the right positions, transform their sizes, once they are located at

these positions, and finally to draw the visual shape, which is a sphere. Next to

position of nodes, texts are located, repositioned and rotated using the billboard

212 M.Trier

convenience class of J3D, which rotates the text in a way, that it always faces the

view (i.e. the observing user). Finally, the actual text is instantiated. Similar to the

nodes of the people network, the edges are drawn. Here, the problem arises, that

the spatial orientation of the edge is depending on the position of the nodes. Fi-

nally, a cylinder shape is instantiated and the complete graph is drawn on the can-

vas.

Figure 83: The conceptual Design Specification of the J3D Renderer.

A special issue for dynamically rendering the extracted data of the people net-

work’s communication is the restriction, that the structure can only be modified in

predetermined ways, once it has been drawn on the canvas (status ‘live’). For ex-

ample, objects can not be added to the drawing shown on a canvas, but they al-

ways have to be defined a priori. This means they initially need to exist before the

Visualization of virtual Knowledge Communities 213

3D structure is shown. The only option for modifications to existing objects is en-

abled by turning on capabilities. Sample capabilities are transformations, geome-

try coordinates, or shape colors. Another complication as compared to 2D graph-

ics is the management of light sourcing, which has to be set up correctly in order

to see the structure.

To handle these issues, the following solutions have been implemented. An ele-

mentary scene graph is generated only once (upfront). The required nodes and

edges are created and cached in the beginning. Text labels are only generated

when requested by the user in order to save processing time. A scene graph can be

updated in dynamic intervals (using the Updater class). To rearrange the network

structure according to the user’s request, the geometrical vectors are not updated,

but the transformations (of edge length or node size). The graph can be rotated and

translated by using a local coordinate system. Zoom is implemented via manipula-

tion of the view transformation. Using the billboard feature of J3D, the text labels

are always facing the user. Further features are the automatic zoom on the graph

and the automatic centralization in the user’s 3D canvas. Further, objects can be

picked, i.e. to view author properties. The problem of not being able to alter the

scenegraph during the visualization is affecting the ability to simply draw new

nodes into the graph as they appear. This is a very important element of time-

based visualization, as the control component constantly adds new nodes which

‘enter’ the conversation. This shortcoming has been overcome by managing the

visibility of scene graph objects.

Summarizing, the J3D renderer provides text labels for edges and nodes, frame

based rendering, mouse-controlled navigation, changes in colors, texts, and visi-

bility, dynamic alteration of edges and nodes, and auto-zoom to optimize the view

on the graph, which is requiring the determination of the optical center of all

nodes.

8.4.2 Animated Longitudinal Visualization and Analysis

Card et al. (1999:30) emphasize that changes over time can also be used to encode

information. Following this idea, a special visualization feature has been devel-

oped: the time-related filter. It allows for observing, how the network actually

evolves over time. The observer can see how clusters are emerging and connecting

with other clusters, or who has been the initial nucleus of a cluster.

The underlying technique is the implementation of a time filter in the backend and

the according addition of a time controller to the layout algorithm. The time filter

allows for generating a set of authors, which are existing in a discourse in a speci-

fied time period.

The time-related animation can either be message driven or time driven. Message

driven animation simply includes the next message in the graph and the layout

component adds it to the visualization. The backend supports the fetching of the

214 M.Trier

very next message in a discourse but also the fetching of a larger bundle of ‘next’

messages in order to allow for higher speeds of animation (also cf. Figure 74 in

chapter 8.3.4). This is necessary, if for example an instant messaging discourse is

visualized, where there are thousands of messages between a limited set of exist-

ing authors. True time driven animation on the other hand adds the next time

frame (from hours to weeks, depending on the overall duration of the discourse

and the speed selected by the user) and fetches the nodes and relations which have

been added in that time period to add them to the graph.

In principle, both types of animation yield in the same result. However, message-

related animation does not visually differentiate periods of large volume increases

and periods of low growth, hence making it impossible to evaluate velocity. On

the other hand, true time related visualization does often not show a meaningful

addition of new messages as multiple messages are added in periods of fast vol-

ume increase.

A third alternative for time driven visualization is time-window-based animation.

This type actually moves a window of a constant size through the sequence of

messages. The result is a visualization of the changes in the network as old struc-

tures which fall out of the time-window again are either faded out or completely

disappear. This helps to emphasize the actual current activity and the added con-

tent.

Technically, the backend’s time filter receives the active time period from the

frontend and retrieves the active nodes and relationships. Here, certain design is-

sues have to be defined - for example, how to display a relation, where the answer

is outside the active period. Here, either both participating authors can be set as

inactive or the one which posted his message can be active and the reply does not

exist or both are treated as active.

Another important design issue is the FR Layout algorithm. It needs to be adapted

as the original algorithm of Fruchterman and Reingold does not take into account

an addition of nodes during the calculation of positions. In the developed layout

algorithm, new nodes come in at random positions and are instantly taken into

consideration with their repulsive and attractive forces. Special attention has been

put to the visual comprehension of the animation. For example, the position of

singular objects should be less sticky then that of larger structures in the current

network. This results in a physical impression of inert large molecules and rather

active and quick new nodes.

Next, the rendering process needs to take into account the temporal addition of

nodes and edges. The 2D renderer can handle this requirement very easily, but the

J3D renderer generates the problem, that its scene graph structures can not be al-

tered during the visualization of the graph. This is why all nodes need to be in-

cluded in the layout, yet they are rendered invisible before they actually ‘appear’

in the discourse.

Visualization of virtual Knowledge Communities 215

The resulting animation is illustrated in the Figure 84. It shows screenshots of the

evolution of an electronic communication network of a corporate instant messag-

ing discourse. From initially unconnected employees, two clusters and a stable

connection between them emerge. In the presented animation, the unconnected

nodes have their initial positions although they have not yet become a part of the

structure. This means that the inactive nodes are not hidden.

Figure 84: Screenshots of the Evolution of an electronic Communication Network.

8.4.3 Semi-manual Content Coding and Analysis

The domain of social network visualization and analysis is the observation and

evaluation of structural patterns. However, in Communities of Practice, not only

the neutral structural patterns are of importance but also the content which has

been discussed in the network. One way to work closer towards a content oriented

analysis is semi-manual content coding using certain coding schemes. This tech-

nique is frequently used by researchers to analyze the structure of communication.

The analyst goes through the complete data set and assigns an according type out

of a predefined set of alternative categories. For this process, the application al-

lows for using a provided Java-based table to add property information. This can

include information about authors’ affiliation or hierarchical status, or also the ap-

plication of a content coding scheme.

Although this semi-manual technique does in its original intention not allow for

identifying hot topics53, it yields in more insight about the actual culture and ap-

plication of the discourse. A sample coding scheme is consisting of elements like

quick answer, document transfer, or socializing communication. For example, if

two persons have a private conversation about their hobbies, this could be re-

garded as a socializing communication. Figure 85 shows a typical scheme for a

classification of contents of a virtual discourse as suggested by Muller et al.

(2003) and Isaacs et al. (2002). It contains 17 categories of communication.

53 This could be done by going through the messages and manually assign a limited and

predefined set of ten to twenty keywords.

216 M.Trier

The information about communication purposes can subsequently be utilized by

the application to indicate different types of contents with different colors, node

sizes, or rings. Alternatively, only the subset of a communication network could

be visualized, which contains the according content coding category. A sample

study could show a subnet which shows only socializing relationships between

employees and compare it with a subnet for document transfer. By this it could for

instance be found out if socializing in networks is a barrier for efficient knowledge

work or if it finally leads to other types of communication and hence supports the

subsequent transfer of knowledge. If the two networks are involving the same au-

thors, then obviously these authors first communicated to socialize before they

started knowledge exchange.

Purposes Description Domain related to

Availability check Asking whether the receiver of IM is available to

start a conversation

Organization

Follow-up Follow-up of previous IM conversations or email

messages

Organization

Ftf meeting coordination Arranging face-to-face formal meetings Organization

Informal meeting coor-

dination

Arranging face-to-face informal meetings Organization

Leave memos Leaving memos without checking conversation

availability

Organization

Media switch Switching communication medium in the middle

of conversation

Organization

Third party call Asking the phone number or availability of a third

party whose presence the IM receiver can check

Organization

Announcement Broadcasting to several IM receivers Organization

Phone call arrangement Arranging phone calls by checking the availability

of the IM receiver

Organization

Information sharing Sharing information related to task completion Knowledge Work

Discussion Discussing issues or problems to complete tasks Knowledge Work

Document transfer Sending and receiving documents Knowledge Work

Quick Q/A Asking a simple question and answering it Knowledge Work

Simple request Asking for a favor to do a simple thing Knowledge Work

Problem solving Discussing a problem in depth and find solutions Knowledge Work

Socializing Having a conversation about what is not related to

work

Socializing

Others Other categories which cannot be grouped into

those described above

Other

Figure 85: A possible Classification of contents in a corporate electronic Instant Messaging

Discourse. Source: cf. Muller et al. (2003) and Isaacs et al. (2002)

8.4.4 Automated Shallow Text Analysis

Next to assigning content types to messages, real topic analysis can be pursued.

The software approach of this book started with shallow text analysis. However, in

later stages more sophisticated procedures can be developed. According to the co-

occurrence of terms in documents, a relation can be computed, indicating the rela-

Visualization of virtual Knowledge Communities 217

tive vicinity of the two terms (or in other words the strength of the relation). Fi-

nally a neural net algorithm, like a Hopfield Network could be added to create a

hierarchy of the concept relations.

The shallow text analysis approach applied for the software analyzes all messages

and their subjects and reads in the full text. This text is then segregated into lists of

words. Afterwards, a ‘stop word’ list is applied, which includes up to 1,000 com-

mon function (or non-semantic bearing) words, such as on, in, at, this, there, etc.

and „pure'' verbs (words which are verbs only), e.g., calculate, articulate, teach,

listen, etc. This list helps to delete high-frequency words that are too general to be

useful in representing document content. Then, a stemming algorithm can identify

the word stem for each remaining word. Sample suffixes include: ive, ion, tion, en,

ions, ications, ens, th, ieth, ly, ing, ings, ed, est, er, ers, s, es, ies, ness, iness, and

's. The resulting keywords are counted for each contribution and stored in decreas-

ing number of occurrence. This procedure yields the most important words for

each subject and each message. By summarizing all messages of an author, the

most relevant terms for that author can thus be found.

A sample approach to utilize keyword analysis in a visualization is shown in

Figure 86. The left-hand side shows the sample keywords, which have been ex-

tracted for the authors S, F, Y, and H (anonymized) in decreasing order of occur-

rence. For example, the message S sent to F dealt with the keywords

‘Socket_select’ (a PHP command), ‘crashes’, ‘http’, ‘net’, and ‘win’. If these

keywords are occurring on both sides of a communication relation (i.e. also in the

answer of F), then they are symmetrical and are placed at the center of the edge

between S and F. Sometimes only one author uses a keyword (e.g. ‘id’ written by

author F). Then this word is located near the author.

A subsequent feature which can be developed after the introduction of keyword

analysis is a search functionality for the people network. All identified terms are

shown in a list or a search term can be entered (like in Figure 86 at the top-right

corner) and the found relations and nodes are marked using a color.

218 M.Trier

Figure 86: Approaches for Keyword Visualization and Analysis.

There are many interesting subsequent features based on keyword analysis possi-

ble. They are part of the measurement approach introduced in chapter 7.2. For ex-

ample, a similarity between two authors can be calculated by looking for their

keyword overlap. Furthermore, the variance of the keywords of the community

can be measured to find out if the discussion limits its initially usually very gen-

eral focus to more specific analyzes.

8.4.5 Implementing and extending Measurements

A final implemented framework is the measurement system developed in section

7.2. Here, in a first stage selected measurements are implemented into the soft-

ware for demonstrating the value of such evaluation functionality (cf. Figure 87).

The actual measurement takes place in the backend. The discourse class is aug-

mented with methods for measuring discourse metrics. These can then be accessed

by the frontend. It has to be differentiated between dynamic properties and static

(or non-dynamic) properties. They differ in the aspect, that dynamic properties are

evaluated for every change in the network, for example when a time driven anima-

tion is played. Examples are messages sent by an author or the relationship

strength. The calculation of static measures is either too resource intensive or the

recalculation of the measure for each change in the visualization is not meaningful

for the analysis. This type of property is thus only updated upon request and re-

mains unchanged during animations. Examples are indirect contacts of an author

or his membership in the core group.

Fopen

http

Fopen

http

Localhost

Receive

Send

Localhost

Receive

Param

id bugs net

Socket_select

Crashes

http

net

win

id fopen

Search:

S > F:

Socket_select crashes http net win

F > S:

Socket_select crashes http net id

Y > -

Fopen http id bugs net

H > Y

Fopen http localhost receive send

S > Y

Fopen http localhost receive param

S > Y

Fopen http localhost receive param

Visualization of virtual Knowledge Communities 219

Figure 87: Two prototypical stages of the Measurement Concept implementation. Left:

Measures are sorted by Entities, Discourse and Author. Right: The Measurement System

Structure has been adopted (parts are invisible).

The implemented measurement system allows creating a first insight about the

quantitative structures, the probability of employing Social Capital benefits, and

knowledge processes. However, it has to be noted, that during the development, it

became clear, that such a measurement system is a big research challenge which

requires further development efforts (which were far beyond the initial scope of

this book’s software approach).

After the implementation of the actual measures, further issues arise. Moderators

need to export the generated evaluations and visualizations. Especially for the

metrics, an export into useful formats like CSV (comma separated value) or even

XML is necessary. Further, an important question is how to make several meas-

urements comparable in order to identify trends. This could even lead to a second

database, which systematically stores the analysis or to an external analysis com-

ponent which is activated and then runs several analysis routines to prepare an ex-

tensive report.

Network Evaluation:

Structural Properties:

Discourse:

Variance of #contacts: 2.4

Variance of relationship strength: 248.5

Coregroup size (#): 46

Coregroup share (%): 66.6

Author (picked):

Elapsed time since last activity(d): 17

Relationship (picked):

Relaltionship strength: 4

Social Capital Properties:

Discourse:

avg. Reciprocity(%): 70

Density(%): 2.13

# Isolates: 12

Coregroup Density(%): 15.3

avg. Relationship strength: 4.2

Author (picked):

avg. Obligation(%): -70.3

avg. Relationship strength: 7.1

#Coregroup contacts: 23

Pulsetaker indicator: 4.1

Broker indicator: 16

direct contacts (Hub-indicator): 29

Relationship (picked):

# mutual contacts of A&B: 2

Property similarity (%) 10

Reciprocity(%) 50%

Knowledge Processes:

Discourse:

Assistance (#A): 23

avg.#Answers/Request: 3.2

Network Evaluation:

Structural Properties:

Discourse:

Variance of #contacts: 2.4

Variance of relationship strength: 248.5

Coregroup size (#): 46

Coregroup share (%): 66.6

Author (picked):

Elapsed time since last activity(d): 17

Relationship (picked):

Relaltionship strength: 4

Social Capital Properties:

Discourse:

avg. Reciprocity(%): 70

Density(%): 2.13

# Isolates: 12

Coregroup Density(%): 15.3

avg. Relationship strength: 4.2

Author (picked):

avg. Obligation(%): -70.3

avg. Relationship strength: 7.1

#Coregroup contacts: 23

Pulsetaker indicator: 4.1

Broker indicator: 16

direct contacts (Hub-indicator): 29

Relationship (picked):

# mutual contacts of A&B: 2

Property similarity (%) 10

Reciprocity(%) 50%

Knowledge Processes:

Discourse:

Assistance (#A): 23

avg.#Answers/Request: 3.2

220 M.Trier

A second interesting issue is the development of a measurement approach for dy-

namic properties. As just mentioned, the dynamics of a people network are highly

interesting for moderators. They want to know, where new hot topics and active

authors are emerging in their network. Although the introduced system proposes

measures for evaluating network growth (i.e. density trend, hot topic, posting fre-

quency trend), much more experiments need to be carried out in this segment of

measurement in order to generate good insights about the longitudinal evolution.

Following the idea of combining visualization and measurement, the measures can

be connected to visualizations. The property approach of the application partially

implements this idea. However, ideally, most measures should be augmented by

color-coding or node-size coding to give a more illustrative impression to the user.

This can be combined with temporal evolvement to highlight growing areas with

colored backgrounds or dying conversations by fading out colors.

If measurement and visualization are combined, then, next to allowing visual rep-

resentation of the measures in the graph, this should consequently lead to a true

cockpit with visual evaluations for allowing quick insights into the current situa-

tion of the network in order to support the management of Knowledge Communi-

ties. The introduction of a conceptual development for this issue will be reserved

for the outlook in chapter 10.

Before that, the next chapter will illustrate the application of the software ap-

proach using several case studies.

Visualization of virtual Knowledge Communities 221

9 Case Studies for IT-supported Community

Visualization and Analysis

After the developed technical approaches have been introduced and discussed, the

resulting software needs to be tested in several case studies in order to demon-

strate its ability to support transparency and subsequently evaluation of virtual ex-

pert groups. Suitable datasets can be found in open discussions, of which some

may be moderated by a commercial entity. Examples are product discussion

groups, where customers share knowledge about complex software or hardware

products. This work will first analyze a non-commercial newsgroup called Slash-

dot.org which informally discusses latest news. A further example, which will be

briefly introduced (anonymized) is an Asian manufacturing company, which used

instant messaging to coordinate their knowledge workers. Finally, a Java Devel-

opers and Help Forum are examined. These cases will be utilized to illustrate the

main features of the software and to emphasize the insights achieved when com-

munities are visualized and evaluated. Finally, a first emerging procedure for IT-

supported community analysis is assembled and documented in chapter 9.4.

9.1 Case Public Slashdot Discussion Group

The first case for testing and applying the software solution is a sample dataset of

a discussion in the public news oriented community board Slashdot. This set is

kept small to allow for an easy demonstration of insights. The communications

between authors have been collected via automatically connecting to the website

that hosts the conversation. Slashdot archives are stored using the hierarchical

paradigm which is translated into the generic database structure via the connector.

The software-specific properties which can be derived are maintained. In Slashdot,

the most interesting individual property is the evaluation of authors. From this

data the author network as shown in Figure 88 is being generated. As described in

section 8.4.1.2, the distance between authors gets smaller, the stronger their com-

munication relation is, i.e. the more they communicate with each other.

From the multitude of Slashdot-based discussions, a group is selected, which dis-

cusses the maintainability of Open Source software. It can serve to illustrate the

ability of including information in the social network visualization. From this dis-

cussion only a small sample was taken to illustrate the analysis using various visu-

alizations. It includes 57 authors, which exchanged 103 messages in 6 days. They

formed 77 relations. The low duration of the sampled discussion is represented by

the resulting low average relationship strength of 1.29. The diameter (longest

path) of the network is 7, which implies that a central author would need to pass

222 M.Trier

about three persons until he reaches the periphery of the network. There are no

isolated authors, whose questions where not commented. 36 persons or almost two

thirds of the available members are necessary to create 80 percent of the board’s

volume. This shows the large structural uniformity with an especially active (epi-)

center or a small core group. Given, that this discussion is employing a hierarchi-

cal discussion paradigm, the density is comparatively high with 4.5 percent (i.e.

every 20th possible relation is actually occurring).

In the network’s visualization (Figure 88), the node size has been set to represent

the activity of the authors in terms of sent messages. The node labels show the au-

thor names, the edge labels show the amount of communication between two

nodes. As this is only a small sample of a short period, there has not been much

communication between any two authors and the relationship strength is mostly

one. This implies that there is not much probability for utilizing Social Capital in

this network, as there are no strong relationships. Considering the fact, that this is

rather an ad-hoc network, it is thus obviously not yet a ‘social’ network.

The node color has been set to represent the authors’ evaluations. This yields in

the next insight. The most active authors54 do not need to be the ones with the best

evaluation. In fact, the node representing the anonymous author is actually de-

creasing the overall evaluation of the network, as its evaluation is particularly low

despite its activity. This pattern also shows that authors usually did not appear to

be highly active in this sample. Rather they comment once and then only read.

Figure 88: Viewing a Slashdot Discussion Group.

54 Which in this sample is a category called Anonymous author, which multiple authors can

employ for posting anonymous contributions.

Visualization of virtual Knowledge Communities 223

The node size could also have been set to represent the average amount of com-

munication with any of the author’s contacts, or in other words the average depth

of his relationships. This uncovers, if the author maintains many weak relation-

ships or rather only few but strong relationships. Generally, this network contains

about four strong relationships. Alternatively, setting the node size to decode mes-

sages received would yield in insights about the prominence of authors. The nodes

who received the most attention in terms of communication are bigger in size.

This network can now also be analyzed for important topics. Figure 89 shows the

result of filtering the keywords read, readable, readability, understandable, and

understand. These terms refer to the readability OpenSource source code and its

role for OpenSource Maintenance. Using the same filtering approach, the topic

‘documentation’ has been found to comprise a large subnet in the overall discus-

sion. Such analysis can help to quickly identify special topics and their sub-

networks in the discussion. It further leads to observing the evolvement of such

topics over time.

Figure 89: The Subnet with the Topic ‘Readability’ of OpenSource Code.

9.2 Case Instant Messaging in a Manufacturing Company

The software application has been applied to study the communication in a manu-

facturing company (initially described in Cho et al., 2005). The observed commu-

nication channel has been instant messaging (IM). The IM transcripts of two peo-

ple A and B have been analyzed in depth. A is at level 1 (out of 5) in the corpo-

rate hierarchy. He analyzes market trends, positions the company in the market

based on the analysis of market trends, analyzes the profitability of the company

based on sales records, and develops the Knowledge Management website for the

company. B is at level 2 in the organizational hierarchy and takes the responsibil-

Subnet: read, readable, readability,

understandable, understand

Subnet: read, readable, readability,

understandable, understand

224 M.Trier

ity for managing the ERP (Enterprise Resource Planning) system and for provid-

ing technical support to remote offices.

As the two individuals where the focus of the analysis, the method of ego-centric

networks (or simply ego networks) has been applied for quantitative analysis and

visualization. This case thus shows, how ego-networks can be analyzed using the

software application introduced in the previous chapters. Such an ego-centric net-

work consists of a focal actor, termed ego, a set of alters who have ties with ego,

and the measurement on the ties among ego and these alters. This data is also re-

ferred to as personal network data. Ego networks are used frequently to study the

social support of persons (Wasserman and Faust, 1994).

For the analysis, a special connection to MSN-Messenger has been developed to

automatically scan, extract, and analyze the network structures in over 80,000

communication acts stored in the sampled electronic Instant Messaging archives

of the company under consideration.

In 217 days, A sent 2574 (45 percent) and received 3185 (55 percent) messages.

Comparing this inbound and outbound traffic, there is no clear evidence for A be-

ing either a passive and prominent collector or an active initiator of messages.

The same applies for B. Comparing the traffic volume of both persons is indicat-

ing a clear difference. In 244 days, B sent 39595 and received 38217 messages.

On average, this equals 162 sent messages per day, whereas A only sent 12 mes-

sages daily, illustrating B’s heavy use of IM. With this large volume of messages,

B maintains 139 contacts, which is more than three times as much as A’s 45 con-

tacts. The different scope of the two persons’ networks can be visualized with an

ego-centric social network visualization, rendered by the software application (see

Figure 90). The star pattern, which has been produced by the layout algorithm, is

typical for an ego-centric data set. The network of B intuitively appears much lar-

ger.

Visualization of virtual Knowledge Communities 225

Figure 90: The ego-networks of Person A and B in two different Departments of a Manu-

facturing Company. Also cf. Cho et al. (2005)

In the visualization, property coding has been applied to store additional informa-

tion about the hierarchical level and the location of the authors (also compare

chapter 8.3.3). Looking at the network of person A, the majority of his 45 contacts

are outside his department on the peer and subordinate level. There are no rela-

tionships to external superiors. In his own department, A has only six contacts.

These contacts are almost only to superiors. However, it has to be recognized that

A has no subordinates within the department. Finally, four external contacts are

maintained. B has a completely different pattern. Of his 139 contacts, more than

two thirds are subordinates which work in the same department. There are only

two contacts to his superiors and no contacts to peers. B’s contacts to other de-

partments are at the peer and the subordinate levels. B maintains only two external

contacts.

About two thirds of A’s received message volume were received from persons of

other departments. Analyzing this subset of external messages, one half was sent

by 21 peers in other departments and the other half of the external messages was

received from 14 external subordinate persons. On average, peers sent fewer mes-

sages per person (53) than subordinates (74). This shows that although comprising

only about one third of A’s external contacts, these subordinate members write

messages more frequently to him.

Compared to A, only 9 percent of B’s messages have been written by employees

of other departments. From them, the biggest group is comprised of external sub-

ordinates, which account for 6 percent of the overall communication volume. Like

with A, the subordinates wrote much more frequent than employees on peer level.

Opposite to A, the external share is very small compared to more than 90 percent

Level1

Level2

Level3

Own DepartmentOther DepartmentExternal Organisational

Hierarchy:

Person EGO

Person A Person B

Short Distance = strong Relationship

226 M.Trier

of messages received from B’s own department. It has mainly been generated by

his subordinates, who also wrote much more frequently than external peers and

subordinates. B has received only one percent of his conversation from higher lev-

els in his department. Both respondents have in common, that they have very few

relationships to internal peers and external superiors.

The analysis of A’s IM use in terms of message sending and reception gives some

insight into his networking patterns. To internal superiors, A is a reactive collector

and receiver with received messages outnumbering sent messages with a ratio of

1.6. This is an interesting result, as in his work relation A would be expected to be

the active reporting person and his superior the receiver and collector of his in-

formation. However, given the tendency of IM conversation towards informal and

spontaneous communications, the superiors of A would be more eligible to apply

informal language to his subordinate A, than it would be in the other direction.

This tendency can also be observed in A’s conversation with external people at

lower levels in the organizational hierarchy. Person A received fewer messages

than he sent (ratio 0.8). Hence, A is more an initiator of informal messages to this

group, than a collector. Again, the superior position in the organizational hierar-

chy is the dominant sender of IM.

To complete the discussion of A’s hierarchical exchange ratios, the conversation

with external and internal colleagues at the same hierarchical level needs to be ex-

plored. Here, A is receiving nearly two times as many messages from external

peers as he sends (1,115 vs. 598 messages; ratio: 1.8). Within his department, the

opposite could be observed: he received 114 messages from his peers and sent 174

(ratio: 1.5).

Looking at person B, to internal superiors he is an active sender with a ratio of 1.8.

This is opposite to person A which has been a receiver of IM messages from

higher ranks. This is interesting when assessing the above assumption, that the in-

formal character of IM communication would suggest that there is not much bot-

tom-up communication with superiors as informal communication is not appropri-

ate for this hierarchical relationship as found for A. However, having in mind, that

B is a heavy user and applies IM less for socializing than A (compare Cho et al.

2005), this result could also confirm B’s employment of IM as a means to increase

efficiency in his work relations. Analyzing B’s relations with lower levels in the

hierarchy reveals that there is no dominant direction (B is a receiver with a very

low ratio of 1.01). To external subordinates, B is even sending more than he re-

ceived. These differences can be related to B’s job profile: He is more supporting

and less reporting than A. A is hence more a collector of information from others,

B more a provider of information.

Using the ego-centric networks, the relationship patterns of A and B can also be

analyzed in more detail by looking for strong ties. For this, a threshold can be de-

fined, which separates strong from weak connections. A pragmatic approach is to

Visualization of virtual Knowledge Communities 227

set it at half the strength of the strongest relation (Rmax/2 with Rmax = strongest

relation in the egonet), thus providing a relative view that eliminates the differ-

ences in usage intensities between A and B. Applying this method, A maintains 11

percent or five strong ties (threshold Rmax/2 = 218 interactions) which are re-

sponsible for 45 percent of the overall message volume. The remaining 89 percent

(40 contacts) are weak ties of A. The tight connections include contacts at differ-

ent hierarchical levels within and outside the department: one peer and two subor-

dinates from external departments plus two internal superiors. B maintains 5 per-

cent or seven strong relationships (threshold Rmax/2 = 1147 interactions). They

generate only 30 percent of B’s processed message volume. Still, comparing the

average amount of interaction, B has considerably stronger communication ties

with his contacts than A. If A’s threshold for a strong tie would be applied to B, he

would maintain 48 relationships at the level of A. The pattern also shows that B

has only strong ties to internal subordinates. Generally, he has only very few

dense relationships with external employees and none with his superiors.

To apply a time-based analysis to this case study, the development of hierarchical

relationships overtime is now visualized and discussed. Figure 91 shows A’s and

B’s egonets during 6 sampled months. Throughout the months, new contacts are

added continuously. Close contacts with strong relationships move to the center.

Figure 91: The Evolution of A’s and B’s ego-centered Network (ungrouped, two month

between each instance of egonet).

Applying the hierarchical and locational categories, Figure 91 shows the devel-

opment of both respondents’ communication network. Within 6 months, A added

26 new contacts to his 17 existing contacts (250 percent growth): 4 external col-

2003-10-01 2003-12-01 2004-02-01 2004-04-01

Person B

2004-02-01 2004-04-01 2004-06-01 2004-08-01

Person A

228 M.Trier

laborators, 4 internal superior, 7 external peers and 11 external superiors. The col-

ors also illustrate that A has developed from internal to external contacts and used

IM for creating cross-departmental relationships to geographically separated loca-

tions. Looking at the job of A, which is to accumulate and analyze information,

this behavior seems to support his work.

B initially started with 35 contacts, and added 1 internal superior, 48 internal sub-

ordinate, 15 external subordinate, 18 external peers, and 2 external collaborators.

In sum, this accumulates to 84 new contacts which is more than 300 percent

growth. However, his primary growth came from developing relationships to his

internal colleagues on lower levels. As he is managing the ERP-system of the

company, this could imply that he is organizing his group using Instant Messag-

ing.

The continuous increase in relationship strength of A and B to colleagues across

hierarchies and locations is depicted in Figure 92.

Figure 92: Development and Increase in Relationship Strength for Persons A and B over 6

Months.

The two ego-networks have also been examined using content analysis (compare

for Cho et al., 2005). This gives many insights into the purpose of applying instant

messaging and the role of Knowledge Work. The content coding scheme intro-

duced in chapter 8.4.3 has been applied. Looking only at the broad domains know-

ledge work, socializing, and (self-) organization, it can be seen, that both in A’s

and B’s electronic communication, messages related to Knowledge Work com-

prise the biggest share (50 percent for A and 67 percent for B). Socializing which

can be interpreted as a means of maintaining and improving relationships to foster

Knowledge Work is also a considerable part of the respondent’s communication

(33 percent for A, 8 percent for B). The different amount of socialization could be

attributed to the fact, that A is more a ‘Collector’ who needs to establish good

connections to draw resources from his network, whereas B is more a ‘Provider’,

100

200

300

400

500

600

700

1234567

A's ext coll

A's int sup

A's ext sup

A's ext peer

200

400

600

800

1000

1200

1234567

B's int sup

B's int sub

B's ext sub

B's ext peer

B's ext coll

Visualization of virtual Knowledge Communities 229

which does not require him to socialize in order to improve his relations. Looking

at the socializing messages reveals that most socializing was actually to check the

results of informal meetings, or send seasonal greetings. This application of con-

tent coding is looking only at broad types of communication. This could be en-

riched by a detailed keyword analysis. However, as most of the conversation of

this case was not in English (Asian company), this was not possible with the cur-

rent software solution.

9.3 Case Public Java Developer Forum

The next case includes a dataset of 10 days (02/04/2005 until 02/13/2005) of dis-

cussion activity in the Java developer forum comp.java.lang.help. During this pe-

riod 161 authors where involved and exchanged 400 messages. Between the 161

authors 183 relations where formed. The longest path between two authors (Di-

ameter) had a length of eleven edges. At the first sight, there are no separated clus-

ters. In the group, 17 persons are isolates (ca. 10 percent). The coregroup includes

81 people, which means, that more than 50 percent of the network’s authors are

responsible for generating 80 percent of the message volume. This indicates that

the network activity is very distributed across the group and there is no small epi-

center of activity.

Figure 93: Complete Network (left) and filtered Sub-network where the terms Call, Calls,

and Constructor occur.

This is obviously due to the fact, that the sampling period is too short to build a

smaller set of core members. The same implication can be derived by looking at

the average relationship strength of 1.81. Thus, only very few messages have been

exchanged to form the relationships. However, the graph shows six strong rela-

Subnet search terms: call,calls,constructorComplete Network

Discussion board: comp.java.lang.help,

10 days: 2005-02-04Æ2005-02-13

161 authors, 400 messages, 183 relations, Diameter 11, 17 Isolates,

coregroup size 81, Average relationship strength 1.81, Density 1.3%

230 M.Trier

tionships (e.g. between node 29 and 40 or between 15 and 41). The overall density

is only 1.3 percent, which shows, that almost all of the possible connections be-

tween authors are not yet realized. This effect can be attributed to the hierarchical

structure of a discussion board in contrast to a peer-to-peer based structure, where

direct relationships are much more important for communication.

In the complete network, there are about seven to eight very active authors, which

moved to the center of the graph. They participate in the strongest relationships in

the sample. On the other hand, they are not very well connected to each other.

This could indicate that they discuss about rather unrelated items and receive their

answers not from each other, but from otherwise not active experts. The sending

activity is indicated by the node size. Node 15 and node 147 were most active. For

example, node 147 sent 17 messages to his 12 direct contacts and received 8 mes-

sages; node 15 has 10 contacts, sent 17 messages and received 19.

Author 119 shows an interesting pattern. He has eighteen connections to otherwise

peripheral contacts. However, the person has sent only two messages. To illustrate

what happened, the node color has been set to represent the received messages.

Orange color shows that author 119 invoked a very impressive feedback (22 mes-

sages) given his low sending activity. Social Network Analysis interprets this po-

sition as ‚prominence’ in the network, representing the attention an author has

gained from others. The question must have been unrelated, as the answers came

by new authors, which previously have not acted in the sampled network. If send-

ing and receiving activity are added, node 15 was most active. The according per-

son participates in the strongest relation, which has emerged between authors 15

and 41. Their dialog includes 15 messages (cf. Figure 94). As two initial requests,

two messages directed to a third person, and one self-related message do not qual-

ify for adding to the relationship strength, the overall communication relation has

a strength of 10. This number thus represents the actual communication, where

one person reacted on the other person’s assertion – or in other words, where a

real information exchange took place. The undirected relationship in the case does

not directly show who is putting more effort into this relationship. Author 15 con-

tributes 9 messages and author 41 6 messages. This looks like author 15 is the pri-

mary driver. However, in terms of strengthening the relationship, author 15 in-

creases the strength by 4 and author 41 by 6 to arrive at the overall relationship

strength of 10. This is due to the fact that an initial request is not adding to the re-

lationship. Further two of the postings of author 15 are directed to third persons.

This shows, although author 15 is writing more messages, the relationship is pri-

mary built by author 41. Thus, quantitatively, this relationship is not completely

symmetrical. Rather author 15 is accumulating some commitment to help author

41 in the future. The notion of commitment becomes even clearer, if the contents

are analyzed. User 15 is only describing problems and author 41 is providing solu-

tions. Obviously user 41 is the expert in this relationship and user 15 the novice.

Looking at the complete network, user 15 seems to be a person, who quickly asks

and draws a lot of resources from the Virtual Community.

Visualization of virtual Knowledge Communities 231

Discussion board: comp.java.lang.help, 10 days: 2005-02-04Æ2005-02-13

Dialog Sequence between 41(Id: 1339) and 15 (Id: 1313) between 2005-02-07 and 02-12

MessID Sent_by

_AuthID Answer to

MsgID Relationsh.

Strength Content Description Content Type

111169 1313 --- Initial

Request Needs help for connecting

to a database with tomcat Problem Description

111216 1339 111169 1 Discusses issues to clarify

the problem a

ain Problem Description,

Solution

111390 1313 111216 2 shows a partial solution

from others and thanks Problem Description,

Solution, Socializin

111249 1313 --- Initial

Request asks for the differentiation

of instantiating,

implementig and

subclassin

Problem Description

111389 1339 111249 3 reassurance of initial

assumptions, adds to the

definitions

Feedback, Problem Solution

111421 1313 111389 4 thanks and asks a related

topic: implements

'implement' all methods?

Soclializing, new Problem

Description

111482 1313 111421 self clarifies mutual

understanding about a

detail of his request

Clarification

111461 1339 111421 5 Answers 'yes' and gives

examples Problem Solution

111476 1313 111461 6 comments and thanks Socialization, Clarification

111192 1313 111175 directed to

a 3rd

person

Tried a JDBC-driver related

solution from a third author -

does not work

Solution Application,

Feedback

111229 1339 111192 7 corrects the initial solution Solution Correction,

Feedback

111539 1313 111531 directed to

a 3rd

person

MySQL driver selection -

solution feedback and

request extension

Clarification

111551 1339 111539 8 gives possible way to solve

the problem - bug fix Feedback, Problem Solution

111553 1313 111551 9 Feedback and Thanks Feedback, Socializing

111554 1339 111553 10 solution extension Solution

Figure 94: The detailed Dialog of the strongest Relationship in the Sample

(between Authors 41 and 15).

Another interesting feature for the analysis is the keyword filter. The right hand

side of Figure 93 illustrates the result of searching the network for the terms call,

calls, and constructor. A sub-network emerges, which comprises the authors and

messages, where these keywords occurred. The rings illustrate the amount of sent

messages. The authors which discussed about these terms can be found immedi-

ately. The five users 1, 44, 143, 115, and 75 can be identified. If the rings are set

to represent the received messages, author 1 is highlighted. He obviously is the

central figure around the search terms. Looking up this author’s contributions

yields in a series with postings with the subject: ‚calls to super must be first state-

232 M.Trier

ment in constructor’. This shows how keyword search can help to quickly find

sub-groups of people which discuss a special issue.

9.4 Conclusion

The previous cases applied the software solution for visualization of knowledge

communities to the analysis of real-world communication networks. Eventually,

the analyses of the cases followed a general procedure, which is influenced by the

available features and emphasis of the software. It is summarized in Figure 95,

which simultaneously suggests a first methodological approach towards a proce-

dure for IT-supported analysis of electronic communication networks. It starts by

interpreting the basic structural and social measures, like message volume or aver-

age relationship strength and density. Here, the general structural uniformity of the

network can be analyzed by comparing the core group size to the size of the over-

all network to see how centralized the epicenter of activity actually is. After study-

ing the numbers, the visualization of the network gets configured to achieve in-

sights into the structural and social patterns. This is done by highlighting impor-

tant structural patterns, active and prominent authors, individual properties (like

author evaluations, hierarchical levels or organizational affiliations), or strong re-

lations and their characteristics (symmetry and obligation). In the next stage, the

contents and keywords are analyzed. This yields in insights about the keywords of

authors, or keywords which are very important in the network.

Figure 95: Towards a Procedure for IT-supported Analysis of Virtual Communication

Networks.

Egonets

and

Growth

Keyword and

Content

Analysis

Visual Structural and

Social Network

Analysis

Structural

& Social

Measure-

ments

1. Analyze structural properties (message volume, number of authors,…)

3. Analyze visualization to visually identify main hubs, brigdes (brokers) (how many?)

6. Analyze importance of authors by identifying most active authors and most prominent

5. Identify and analyze strongest relationships (symmetry, obligation)

2. Analyze uniformity by comparing coregroup size to overall size

8. Visualize and understand keywords of authors

9. Filter for sets of similar keywords which yield in a large subset to identify relevant topics

7. Visualize and understand individually stored properties (e.g. author evaluation)

11. Filter and understand most important ego-networks

4. Filter out weak relationships and authors with low activity to reveal the core structure

12. Observe the evolution over time to identify special structural or time related patterns

10. Analyze meaningful dialogs in the strongest relationships

Visualization of virtual Knowledge Communities 233

The simple keyword study can be extended by observing the actual contents, e.g.

by tracing some meaningful dialogs of strong relationships. Together with the pre-

viously analyzed properties of the relationships this can yield insights about

knowledge processes. The final step takes a closer look at individual subnets of

authors (ego-centric networks) and their properties. Finally, the growth and evolu-

tion of the network is studied using the longitudinal animation.

The four stages shown in Figure 95 are obviously related to the four measurement

domains for the analysis of electronic communication networks, namely structural,

social network, knowledge process and growth measures, as introduced in section

7.2. It has to be noted that more research is necessary for developing applicable

measures (and related visualizations) for the observation and evaluation of knowl-

edge processes and growth. Another important implication is that in large net-

works, the filtering approaches take a very central role to reduce the overall net-

work to meaningful subnets which allow observation, measurement, and evalua-

tion. This can include filtering for important authors, relations, time sequences, or

topics.

Next to these future development needs, it is also necessary to note, that actual

measurement is still difficult, especially if it is to prepare the qualitative evalua-

tion of working versus poor structures. This is mainly due to some conflicting op-

tions for interpretation. It is simply hard to tell, whether a structure is poor. Here,

some first list of objective criteria needs to be discovered.

For example, if over 6 months A sent fewer messages than he received, does this

indicate his obligation to write more in the future or does this simply represent his

normal network position as a collector of information? Further, interpretation of

measures strongly corresponds to the type of network. For example, in a discus-

sion board the necessary density is usually lower than in an e-mail network, as all

communication is centrally visible, so that not every new person needs to get into

contact with the expert and on average, the number of ‘necessary’ relations is

therefore lower. Another example is the actual establishment of relationships. Al-

though frequent communication forms this relationship, such frequent communi-

cation in a community with virtually unlimited members does rarely happen.

There seems to be the primacy of topic-orientation. The users do not look for gen-

erating relationships but for helping others out if their topic of expertise is re-

quested. However, the second case showing the actual long term corporate appli-

cation proves that strong relationships are being built to a limited set of persons.

Finally, when looking at all the results of the cases, an indirect but important issue

needs to be discussed. It is the immediate danger for privacy issues, which is

sometimes causing criticism. Here it has to be checked, if legal rights of the stud-

ied individuals are affected. This is especially the case in peer-to-peer based net-

works (e.g. e-mail or instant messaging), where the individual person stores his

messages and thus is interested in a high degree of privacy. In discussion boards,

234 M.Trier

this issue is elevated to the group level. Every person contributing to the group

agrees on the public character of his posting within that group. However, it has to

be ensured, that the group as a unit enjoys a certain degree of privacy. This issue

can be dealt with by only working with anonymized authors. However, in some

corporate applications of this analysis, it might be the objective to identify a spe-

cial expert person. Here, careful considerations have to be carried out to ensure a

satisfactory protection of private data and to avoid a detrimental behavior of the

group’s members.

Visualization of virtual Knowledge Communities 235

10 Outlook: Towards IT-supported People Network

Management

The approach towards the IT-supported visualization, analysis, measurement, and

evaluation is just a step towards a more professional application of virtual com-

munication networks. They have various advantages for corporations, which have

yet to be discovered and tested. For example, such communication channels allow

for permanent expert groups which can replace teams that only meet once in a

while. The Virtual Community can asynchronously and on-demand work on top-

ics, even if the people are geographically distributed (which is already given, if

they work in different departments). Further advantages are threaded and system-

atical dialogs in topic areas, which are automatically protocolled in a persistent

and accessible communication archive together with related documents. This re-

sults in the ability to host much larger groups in the Virtual Community as it

would have been possible in a face-to-face community, which allows for integrat-

ing more expertise.

All these advantages together with the ability to keep track of the development

and evolution of the virtual communication network using IT-supported monitor-

ing can eventually lead to People Network Management (PNM) supported by So-

cial Network Intelligence Software (SNIS), which is uncovering the complex

structures and their exchange mechanisms to better understand Social Capital and

its access to distributed resources in people networks. Further it enables to under-

stand and support informal communication to utilize it to generate value for the

employees and subsequently for an organization.

Figure 96: A concept for People Network Management in a Virtual Community using

Social Network Intelligence Software.

Execute

Management and

Moderation Tasks in the

lifecycle of the CoP

Monitor and Measure

Group Structure, Contents,

Progress, Development,…

using Measurement

Approach

Plan

Management and

Moderation Tasks in the

lifecycle of the CoP

Allign Plan

Commitment

Stakeholders in

the organization

Reporting

Stakeholders in

the organization

236 M.Trier

The first necessary element of such a People Network Management is the planning

of intended management and moderation tasks for a specific community. This can

be influenced by commitments which the community has to sponsors, service de-

partments, or Knowledge Management strategies. The plan is executed in the ac-

tual task of moderating or managing the group through the stages of the lifecycle

as described in chapter 5.8. Throughout this work a systematical monitoring of the

group’s development can be employed. This activity utilizes the software applica-

tion as introduced in chapter 8. The achieved insights can then be applied to con-

figure the structural properties, described in chapter 5.4 or to adjust management

activities again and the loop can start over (cf. Figure 96). Additionally, a report

documentation can be created to communicate the progress and value to the mem-

bers and to external stakeholders.

Figure 97: A Cockpit for IT-supported People Network Management.

The continuous observation of the group’s structure, its social network evolution,

and knowledge processes can eventually be supported by an actual monitoring

cockpit. An illustration of how such a cockpit for IT-supported management of

Virtual Communities could look like is given in Figure 97. For each of the meas-

urement domains, two dashboard instruments have been selected. For example, in

the structural domain, the manager observes the ratio of active versus registered

users and the ratio of the core group versus all members to analyze participation

and uniformity of the network. If the core group share gets to high, this implies a

closed group in the center, which does not react satisfactorily to external requests.

Visualization of virtual Knowledge Communities 237

By setting his generally expected ratios as normative benchmarks, the moderator

can quickly see if activity decreases or centrality increases more than intended.

Currently, the traffic light is green, meaning that every measure is in the tolerance

zone. In the domain Social Capital formation, the manager observes density,

which is logarithmically scaled and size over time. In the knowledge process do-

main, he has selected the indicators for average relationship strength and average

reciprocity. Here the traffic light is yellow, implying that one of these measures

approaches a threshold limit selected by the manager. The growth dimension dis-

plays a measure for the Average Time Between Messages Sent (ATBMS), which

is a proxy for the frequency or speed of the posting activity. Further, there is a

measure which represents growth. The traffic light implies that one of these fac-

tors is approaching the threshold for an alert.

The previous example of a Cockpit interface shows, what could eventually be de-

veloped using the methodology and the software approach introduced in this book.

With such concepts, the moderation of communities could eventually be advanced

towards a true (people network) management role. Still, it has to be emphasized

again, what already has been discussed in chapter 5.6: the management of net-

works is very different to the conventional understanding of managing. It can not

employ hierarchical authority and has to consider the network itself as an object

for analysis and intervention. Primarily, the social aspects come first, which render

the manager more a facilitator than a decider for the group. This is simultaneously

a very modern approach to managing in a complex environment with autonomous

knowledge workers which derive value from expert networks.

Finally, it can be concluded, that the further development of software applications

which support virtual community monitoring is a rewarding research area. This

coincides with the recent publication of Gartner Group (2004) about the future

growth domains for Emerging Technologies: It defined ten software fields as ‘On

the Rise’. Next to topics like RFID, Augmented Reality, Wikis, Information Ex-

traction, or Truth Verification, software which supports Social Network Analysis

is on this list. This promises much further development in this field during the

next years.

238 M.Trier

11 Literature

(Abecker et al., 2002):

Abecker, A.; Hinkelmann, K.; Maus, H.; Müller, H.J.: Integrationspotenziale

für Geschäftsprozesse und Wissensmanagement (in German). In: Abecker, A.,

Hinkelmann, K., Maus, H., Müller, H.J. (eds.): Geschäftsprozessorientiertes

Wissensmanagement: Effektive Wissensnutzung bei der Planung und Umset-

zung von Geschäftsprozessen (in German). Springer-Verlag, Berlin Heidelberg

New York Tokio, 2002, p. 1-22.

(Albrecht, 1993):

Albrecht, F.: Strategisches Management der Unternehmensressource Wissen

(in German). Berlin, 1993.

(Allee, 2000):

Allee, V.: Knowledge Networks and Communities of Practice. OD Practitioner

Online, Vol. 32, Nr. 4, 2000.

(Ambrozek and Cothrel, 2004):

Ambrozek, J.; Cothrel, J.: Online Communities in Business: Past Progress, Fu-

ture Directions. 7th International Conference on Virtual Communities, July

2004.

(Anderson, 1995):

Anderson, J. R.: Cognitive Psychology and its Implications, New York, 1995.

(APQC, 2001):

APQC: Assessing the Health and Effectiveness of Communities of Practice.

http://old.apqc.org/free/articles/dispArticle.cfm?ProductID=1389. Accessed:

2003-01-20.

(Argyris and Schoen, 1978):

Argyris, C.; Schoen, D.A.: Organizational Learning: A Theory of Action Per-

spective. Reading, Massachussets, 1978.

(Argyris and Schoen, 1996):

Argyris, C.; Schoen, D.: Organizational Learning II, Theory, Method, and

Practice, Reading, 1996.

(Aristotle, 1993):

Aristotle, Analytica Posteriora. First volume, Berlin, 1993.

(Baddeley, 1997):

Baddeley, A.: Human Memory. Theory and Practice. Hove, UK, 1997.

Visualization of virtual Knowledge Communities 239

(Baecker, 1999):

Baecker, D.: Organisation als System (in German). Frankfurt am Main, 1999.

(Barley, 1996):

Barley, S. R.: Technicians in the workplace: Ethnographic evidence for bring-

ing work into organization studies. Administrative Science Quarterly,

41(1996)3, p. 404-441.

(Barnard, 1938):

Barnard, C.I.: The Functions of the Executive. Cambrigde, Massachusetts,

1938.

(Barney, 1991):

Barney, J. B.: Firm resources and sustained competitive advantage. Journal of

Management. 17/1991, p.99-120.

(Bateson, 1972):

Bateson, G.: Steps to an Ecology of Mind. San Francisco, 1972.

(Battista, 1999):

Battista Di, G.: Graph Drawing: Algorithms for the Visualization of graphs.

Upper Saddle River, New Jersey, 1999.

(Bednarz, 1988):

Bednarz, J.: Autopoiesis: The Organizational Closure of Social Systems. Sys-

tems Research, 5(1988)1, p. 57-64.

(Bell, 1973):

Bell, D.: The Coming of Post-industrial Society. A Venture of Social Forecast-

ing, New York, 1973.

(Ben Porath, 1980):

Ben Porath, Y.: The F-Connection: Families, friends, and firms and the organi-

zation of exchange. Population and Development Review, 6/1980, p.1-30.

(Bereiter, 1994)

Bereiter, C.: Implications of postmodernism for science, or, science as progres-

sive discourse. Educational Psychologist, 29(1994), p. 3-12.

(Berge and Collins, 2000):

Berge, Z. L.; Collins, M. P.: Perceptions of e-moderators about their roles and

functions in moderating electronic mailing lists. Distance Education: An Inter-

national Journal, 21(2000)1, p. 81-100.

(Berger and Luckmann, 1966):

Berger, P. L.; Luckmann, T.: The Social Construction of Reality: A Treatise in

the Sociology of Knowledge, New York, 1966.

240 M.Trier

(Bertallanfy, 1950):

Bertallanfy von, L.: An Outline of General Systems Theory. The British Jour-

nal for the Philosophy of Science, 1(1950)2, p.135-165.

(Bertin, 1983):

Bertin, J.: Semiology of Graphics. The University of Wisconsin Press. Madi-

son, 1983.

(Blumer, 1973):

Blumer, H.: Symbolischer Interaktionismus und Ethnomethodologie (in Ger-

man). Reinbek, 1973.

(Bock and Husain, 1952):

Bock, R. D.; Husain, S. Z.: Factors of the tele: a preliminary report. Sociome-

try. (15)1952, pp. 206-219.

(Boje and Dennehy, 1993):

Boje, D. M.; Dennehy, R. F.: Managing in the postmodern world: America’s

revolution against exploitation. Kendall and Hunt, Dubuque, 1993.

http://cbae.nmsu.edu/~dboje/papers/restorying_reengineering_1997.htm. Ac-

cessed: 2004-10-10.

(Bonifacio et al., 2004):

Bonifacio, M.; Camussone, P.; Zini, C.: Managing the KM Trade-Off: Knowl-

edge: Centralization versus Distribution. Journal of Universal Computer Sci-

ence (J.UCS), 10(2004)3.

(Brass, 1985):

Brass, D.: Technology and structuring of jobs: Employee satisfaction, per-

formance, and influence. Organizational Behavior and Human Decision Proc-

esses, 35(1985)2, p.216-240.

(Brown and Duguid, 1991):

Brown, J. S.; Duguid P.: Organizational Learning and Communities-of-

Practice; Toward a Unified View of Working, Learning and Innovation. Or-

ganization Science 2, 1/1991, p.40–57.

(Brown and Duguid, 1998):

Brown, J. S.; Duguid, P.: Organizing Knowledge. In: California Management

Review, 40(1998)73.

(Brown and Gray, 1995):

Brown, J. S.; Gray, E. S.: The People are the Company. In: Fast Company,

November 1995

(Bullinger et al., 1997):

Bullinger, H.-J.; Wörner, K.; Prieto, I.: Wissensmanagement heute. Daten,

Fakten, Trends (in German). Fraunhofer IAO, Stuttgart, 1997.

Visualization of virtual Knowledge Communities 241

(Bullinger, 1998):

Bullinger, H.-J.: Wissensmanagement – Anspruch und Wirklichkeit: Ergebnis-

se einer Unternehmensstudie in Deutschland (in German). In: Information

Management, 1/1998, p.7-23.

(Burrell and Morgan, 1979):

Burrell, G.; Morgan, G.: Sociological Paradigms and Organizational Analysis:

Elements of the Sociology of Corporate Life. Heineman, London, 1979.

(Burt, 1992):

Burt, R. S.: Structural Holes. Cambridge, Massachussets, 1992.

(Cachel and McDermott, 2001):

Cachel, J.; McDermott, R.: Interview between Cachel and McDermott.

http://www.onlinecommunityreport.com/features/mcdermott. Accessed: 2004-

10-10.

(Card et al., 1999):

Card, S.K.; Mackinlay, J.D.; Schneiderman, B.: Information Visualization, Us-

ing Vision to Think. Morgan Kaufmann Publisher, San Francisco, 1999.

(Castells, 2000):

Castells, M.: Materials for an exploratory theory of the network society, British

Journal of Sociology, 5(2000)1, p.5-34.

(Chang et al., 2002):

Chang C.-L.; Chen, D.-Y., Chuang, T.-R.: “Browsing newsgroups with a so-

cial network analyser”; Proc. 6th International Conference on Information

Visualization, IEEE Computer Society Press. London, 2002, p.750-755.

(Cho et al., 2005):

Cho, H.-K.; Trier, M.; Kim, E.: The Use of Instant Messaging in Working Re-

lationship Development: A Case Study. Preliminary Acceptance for Publica-

tion in Journal of Computer-mediated Communication 10(2005)4.

(Christmann-Jacoby and Maas, 1997):

Christmann-Jacoby, H.; Maas, R.: Wissensmanagement im Projektumfeld auf

Basis von Internet-Technologien. In: Information Managment, 3/97, p. 16-26.

(Cohen et al. 1972):

Cohen, M. D.; March, J. G.; Olsen, J. P.: A Garbage Can Model of Organiza-

tional Choice. Administrative Science Quarterly, 17(1972), p.1-25.

(Coleman et al., 1966):

Coleman, J. S.; Katz, E.; Menzel, H.: Medical innovation. New York, 1966.

(Cothrel & Williams, 1999):

Cothrel, J.; Williams, R.: On-line communities: Helping them form and grow.

Journal of Knowledge Management, 3(1999)1, p.54-65.

242 M.Trier

(Cothrel, 2000):

Cothrel, J.P.: Measuring the success of an online community. Strategy & Lead-

ership 28(2000)2, p.17-21, http://www.participate.com/ research

/StrategyLeadership.pdf, Accessed: 2002-11-25.

(Covisint, 2002):

Covisint: About Covisint; Covisint Solutions, 2002, www.covisint.com, Ac-

cessed: 2002-11-09.

(Davenport and Prusak, 1998):

Davenport, T.; Prusak, L.: Working Knowledge. How Organizations Manage

What They Know. Harvard Business School Press, 1998.

(Davenport, 1998):

Davenport, T.: Successful Knowledge Management Projects. In: Sloan Man-

agement Review, 39(1998)2, p.43-58.

(De Geus, 1997):

De Geus, A.P.: The Living Company, Boston, Massachussets, 1997.

(Deal and Kennedy, 1982):

Deal, T. E.; Kennedy, A. A.: Corporate Cultures: The Rites and Rituals of Cor-

porate Life, Reading, Massachussets, 1982.

(Descartes, 1979):

Descartes, R.: Geschichte der Philosophie in Text und Darstellung (in Ger-

man). Volume 5, Rationalismus (in German). In: Specht, R. (Ed.) Stuttgart,

1979, p.99-139.

(Donath et al., 1999):

Donath, J., Karahalios, K., Viegas, F.: Visualizing conversations. Proceedings

of HICSS 32. Hawaii, 1999. http://smg.media.mit.edu/papers/Donath/ Visual-

Conv/VisualizeConv.pdf, Accessed: 2004-03-10.

(Dresdner Bank, 2002):

Dresdner Bank: Branchenreport Maschinenbau (in German). Dezember 2002.

(http://www.dresdner-bank.de/meta/kontakt/01_economic_research/18_bran-

chenservice/ muster.pd, Accessed: 2004-03-01.

(Droege et al., 1993):

Droege, W.; Backhaus, K.; Weiber, R.: Strategien für Investitionsgütermärkte.

Landsberg/Lech, 1993.

(Drucker, 1993):

Drucker P.: Post-capitalist society. New York, 1993.

Visualization of virtual Knowledge Communities 243

(DTI Automotive Unit, 2004):

Automotive Unit of the Department of Trade and Industry: Online Statistics.

http://www.autoindustry.co.uk/statistics/key/production/employees, Accessed

2004-03-10.

(Durkheim, 1977):

Durkheim, E.: Über die Teilung der sozialen Arbeit. Frankfurt am Main, 1977.

(Earl, 2001):

Earl, M.J.: Knowledge Management Strategies: Towards a taxonomy. In: Jour-

nal of Management Information Systems, 18(2001)1, p.215-233.

(Enkel et al., 2002):

Enkel, E.; Heinold, P.; Hofer-Alfeis, J.; Wicki, Y.:The power of communities:

How to build Knowledge Management on a corporate level using a bottom-up

approach. In: Davenport, T., Probst, G. (Eds.), Knowledge Management Case

Book, Erlangen, 2002, p.108-127.

(Erickson and Kellogg, 2000):

Erickson, T. ; Kellogg, W. A.: Social Translucence: An Approach to Designing

Systems That Support Social Processes. ACMTransactions on Computer-

Human Interaction 7(2000)1, p.59–83.

(Erickson and Laff, 2001):

Erickson, T.; Laff, M. R.: The Design of the ‘Babble’ Time-line: A Social

Proxy for Visualizing Group Activity over Time. Extended Abstracts. Proceed-

ings of CHI 2001. ACM Press: New York, 2001.

(Etzioni and Etzioni, 1999):

Etzioni, A.; Etzioni, O.: Face-to-face and computer-mediated communities, a

comparative analysis, The Information Society 15(1999)4, p. 241-248.

(Etzioni, 1967):

Etzioni, A.: Soziologie der Organisation. Juventa, München, 1967.

(European Commission, 2003a):

European Commission, Directorate-General for Research: Key Figures

2003/2004 - Towards a European Research Area - Science, Technology and

Innovation, 2003. ftp://ftp.cordis.lu/pub/indicators/docs/ind_kf0304.pdf, Ac-

cessed: 2004-03-01.

(European Commission, 2003b):

European Commission Directorate-General for Press and Communication: Go-

ing for growth - The economy of the EU. http://www.europa.eu.int/comm/

publications/booklets/move/40/en.pdf, Accessed 2004-03-03.

244 M.Trier

(European Commission, 2004):

European Commission: Connecting Europe at high speed: recent developments

in the electronic communications sector. EU Institutions press releases. 2004-

02-03. http://www.europa.eu.int, Accessed: 2004-03-10.

(Eurostat, 2004):

Eurostat Statistical Office of the European Communities: No title.

http://www.europa.eu.int/comm/eurostat/. Accessed 2004-03-10.

(Ferguson, 1767):

Ferguson, A.: An essay on the history on civil society. Millar and Caddel, Lon-

don, Kincaid and Bell, Edinburgh, S. I1. http://www.socsci.mcmaster.ca./

~econ/ ugcm3ll3 /ferguson/civil.html, Accessed: 2004-10-01.

(Festinger, 1950):

Festinger, L.: Informal Social Communication. Psychological Review,

57(1950), p.271-282.

(Fontaine, 2001):

Fontaine, M.: Research report and toolkit: Understanding, identifying and se-

lecting the roles needed to staff and support communities of practice. Research

Report, IBM Institute for Knowledge Management, Cambridge, USA, 2001.

(Forrester Research, 1999):

Forrester Research: Web Site Success Transcends Hits And Page Views.

http://www.forrester.com/ER/Press/Release/0,1769,197,FF.html, Accessed:

2003-06-21.

(Freeman, 2000):

Freeman L.C.: Visualizing Social Networks. Journal of Social Structure

(1)2000.

(Fruchterman and Reingold, 1991):

Fruchterman, T. M. J.; Reingold, E. M.: Graph Drawing by Force-Directed

Placement, Software - Practice & Experience 21(11)1991, p. 1129-1164.

(Fukuyama, 1995):

Fukuyama, F.: Trust: The Social Virtues and The Creation of Prosperity. New

York: Penguin Books, 1995, p. 26.

(Gartner Group, 2004):

Gartner Group: Hype Cycle for Emerging Technologies. Gartner Research Re-

port, 2004.

(Garton et al., 1997):

Garton, L.; Haythornthwaite, C.; Wellman, B.: Studying Online Social Net-

works. Journal of Computer Mediated Communication, 3(1997)1.

Visualization of virtual Knowledge Communities 245

(Gebert and von Rosenstiel, 1992):

Gebert, D.; Rosenstiel von, L.: Organisationspsychologie: Person und Organi-

sation (in German). Stuttgart et al., 1992.

(Girgensohn et al., 2003):

Girgensohn, A.; Lee, A.; Zhang, J.: Social Browsers for Visualizing Web

Communities. WWW 2003, May 20-24, ACM 2003, Budapest, Hungary.

(Glasersfeld, 1996):

Glasersfeld von, E.: Radikaler Konstruktivismus. Frankfurt am Main, 1996.

(Gongla and Rizzuto, 2001):

Gongla P.; Rizzuto C. R.: Evolving communities of practice: IBM Global Ser-

vices experience. IBM SYSTEMS JOURNAL, 40(2001)4, p.842-862.

(Granovetter, 1973):

Granovetter, M. : The Strength of Weak Ties. American Journal of Sociology,

78(1973), p. 1360-1380.

(Granovetter, 1974):

Granovetter, M. S.: Getting a job. Cambridge, Massachussets, 1974.

(Granovetter, 1985):

Granovetter, M. S.: Economic action and social structure: The problem of em-

beddedness. American Journal of Sociology, 91(1985), p.481-510.

(Grant, 1996):

Grant, R.: Toward a knowledge-based theory of the firm. Strategic Manage-

ment Journal, 17(1996), p.109-123.

(Grochla, 1970):

Grochla, E.: Systemtheorie und Organisationstheorie (in German). In: Grochla,

E. (Ed.): Organisationstheorie (in German). Volume 1, Stuttgart, 1976, p.554-

566.

(Gronau et al., 2003):

Gronau, N.; Palmer,U.; Schulte, K.; Winkler, T.: Modellierung von wissensin-

tensiven Geschäftsprozessen mit der Beschreibungssprache K-Modeler (in

German) In: Reimer, U. Abecker A., Staab S., Stumme G., (eds.): Professio-

nelles Wissensmanagement – Erfahrungen und Visionen (in German). Procee-

dings GI 2003, p.315-322.

(Gueldenberg, 2001):

Wissensmanagement und Wissenscontrolling in lernenden Organisationen: ein

systemtheoretischer Ansatz (in German). Wiesbaden, 2001

(Hahn, 1996):

Hahn, D.: Planung und Kontrolle, Planungs- und Kontrollsysteme, Planungs-

und Kontrollrechnung (in German). Wiesbaden, 1996.

246 M.Trier

(Hanker, 1990):

Hanker, J.: Die strategische Bedeutung der Informatik für Organisationen: In-

dustrieökonomische Grundlagen des strategischen Informationssmanagements

(in German). Stuttgart, 1990.

(Hansen et al., 1999):

Hansen, M. T.; Nohria, N.; Thierney, T.: Wie managen Sie das Wissen in Ih-

rem Unternehmen (in German)? In: Harvard Business Manager, 5/1999, p.85-

96.

(Hansen, 1999):

Hansen, M.T.: The search transfer problem: The role of weak ties in sharing

knowledge across organizational sub-units. Administrative Science Quarterly,

44)(1999), p-82-111.

(Hedberg, 1991):

Hedberg, B.: How organizations learn and unlearn. In: Nystrom, P.C.; Star-

buck, W.H. (Eds.) Handbook of Organizational Design, New York, Oxford,

1981, p.3-27.

(Heidegger, 1986):

Heidegger, M: Sein und Zeit (in German). Tübingen, 1986.

(Heinold, 2000):

Heinold, P.: The power of communities: how to build Knowledge Management

on a corporate level using a bottom-up approach. Online Proceedings of Work-

shop MK2, Munich, 2000. http://www11.in.tum.de/forschung/foren/ mkmk/

proceedings/ community/heinold.ppt, Accessed 2003-08-20.

(Heisig, 2003):

Heisig, P.: Wissensmanagement in industriellen Geschäftsprozessen (in Ger-

man). Industrie Management 3, Gito-Verlag, Berlin, 2003, p. 22-25.

(Hejl, 1984):

Hejl, P.: Towards a Theory of Social Systems: Self-Organization and Self-

Maintenance, Self-Reference and Syn-Reference. In: Ulrich, H.; Probst, G.

(eds.): Self-Organization and Management of Social Systems: Insights, Prom-

ises, Doubts, and Questions. Berlin, 1984, p. 60-78.

(Hemphill and Westie, 1950):

Hemphill, J. K.; Westie, C. M.: The Measurements of Group Dimensions.

Journal of Psychology 29, 1950, p. 325–342.

(Henschel, 2000):

Henschel, A.: Communities of Practice: Plattform für organisationales Lernen

und den Wissenstransfer (in German). Dissertation, St.Gallen, 2000.

Visualization of virtual Knowledge Communities 247

(Hildreth et al., 1999):

Hildreth, P.; Wright, P.; Kimble, C.: Knowledge Management: Are We Miss-

ing Something? In: Brooks, L.; Kimble, C.: Information Systems - The Next

Generation. Proceedings of the 4th UKAIS Conference, York, UK, April 1999,

p.347-356. http://www.cs.york.ac.uk/mis/, Accessed 2003-12-12.

(Hillery, 1955):

Hillery, G. A.: Definitions of Community: Areas of Agreement. Rural Sociol-

ogy, 20(1995), p. 111-123.

(Hinkelmann et al., 2002):

Hinkelmann, K.; Karagiannis, D.; Telesko, R.: PROMOTE – Methodologie

und Werkzeuge für geschäftsprozessorientierten Wissensmanagement (in Ger-

man). In: Abecker, A., Hinkelmann, K., Maus, H., Müller, H.J. (eds.): Ge-

schäftsprozessorientiertes Wissensmanagement: Effektive Wissensnutzung bei

der Planung und Umsetzung von Geschäftsprozessen (in German). Springer-

Verlag, Berlin Heidelberg New York Tokio, 2002, p.65-90.

(Hutchins, 1996):

Hutchins, E.: Cognition in the Wild. Cambridge, Massachussets, 1996.

(IFO, 2001):

Institut fuer Wirtschaftsforschung: Der mittelständische Maschinenbau am

Standort Deutschland – Chancen und Risiken im Zeitalter der Globalisierung

und New Economy (in German). October 2001.

http://www.ifo.de/pls/portal30/docs/FOLDER/IFO_PORTAL/IFO_INSTITUT

/WIR_UEBER_UNS_X _ABOUT_IFO/BEREICHE/STRUKTURWANDEL

_UND_BRANCHEN_SB_/ PROJEKTE_SB/MASCHINENBAU-STUDIE.

PDF, Accessed 2004-03-10.

(Jarvenpaa and Leidner, 1998):

Jarvenpaa, S. L.; Leidner, D. E.: Communication and Trust in Global Virtual

Teams. In Journal of Computer Mediated Communication, 3(1998)4.

(Johnson, 2001):

Johnson, C. M.: A survey of current research on online communities of prac-

tice. Internet and Higher Education 4(2001), p. 45–60.

(Kant, 1980):

Kant, I.: Kritik der reinen Vernunft, Band 1 (in German). Frankfurt, 1980.

(Kantor and Lapsley, 1986):

Kantor, B.; Lapsley, P.: RFC 977: Network News Transfer Protocol - A Pro-

posed Standard for the Stream-Based Transmission of News, 1986.

http://www.w3.org/Protocols/rfc977/ rfc977.html, Accessed 2004-01-12.

248 M.Trier

(Kieser and Kubicek, 1992):

Kieser, A.; Kubicek, H.: Organisation (in German). Berlin, New York, De-

Gruyter. 1992

(Kim, 1993):

Kim, D.H.: The Link between Individual and Organizational Learning. In:

Sloan Management Review, 35(1993)1, p.37-50.

(Kim, 2000):

Kim, A. J.: Community Building on the Web. Peachpit Press: Berkeley, USA,

2000.

(Kirsch, 1969):

Kirsch, E.: Die Unternehmungsziele aus organisationstheoretischer Sicht (in

German). In: ZfbF, 21(1969), p.665-675.

(Kleinhans, 1989):

Kleinhans, A.M.: Wissensverarbeitung im Management (in German). Frank-

furt et al., 1989.

(Kowol and Krohn, 1995):

Kowol, U.; Krohn, W.: Innovationsnetzwerke. Ein Modell der Technikgenese

(in German). In: Halfman, J.; Bechmann, G.; Rammert, W. (eds.): Technik und

Gesellschaft, Jahrbuch 8, Frankfurt a.M., New York, 1995, p.77-104..

(KPMG, 2003):

KPMG: KPMG Knowledge Management Survey – Results Jan 2003.

http://www.knowledgeboard.com/download/1935/ kpmg_km sur-

vey_results_jan_2003.pdf, Accessed: 2004-01-10.

(Krackhardt et al., 1995):

Krackhardt, D.; Blythe, J.; McGrath, C.: KrackPlot 3.0 User's Manual, Carne-

gie-Mellon University, Pittsburgh, 1995.

(Krackhardt, 1991):

Krackhardt, D.: The strength of strong ties: The importance of philos in or-

ganizations. In Nohira, N.; Eccles, R. (Eds.): Organizations and networks:

Theory and practice, Cambridge, MA, 1991, p. 216-239.

(Krallmann et al., 1999):

Krallmann, H.; Frank, H.; Gronau, N.: Systemanalyse im Unternehmen (in

German). 3rd Edition. Oldenbourg Verlag, München, Wien, 1999.

(Krogh and Roos, 1996):

Krogh, G. von; Roos, J.: Five claims on knowing. European Management Jour-

nal 14(1996)4, p. 423-426.

Visualization of virtual Knowledge Communities 249

(Kuehn and Abecker, 1998):

Kuehn, O.; Abecker, A.: Corporate Memories for Knowledge Management in

Industrial Practice. In: Borghoff, U. M.; Pareschi, R. (eds.): Information Tech-

nology for Knowledge Management. Berlin, p. 183-206.

(Lane and Bachmann, 1996):

Lane, C.; Bachmann, R.: The social constitution of trust: supplier relations in

Britain and Germany. Organization studies, 17(1995)3, p.365-395.

(Lane and Lubatkin, 1998):

Lane, P.J.; Lubatkin, M.: Relative absorptive capacity and interorganizational

learning. Strategic Management Journal 19(1998)5, p.461–477.

(Lave and Wenger, 1991):

Lave, J.; Wenger, E.: Situated Learning. Legitimate Peripheral Participation.

Cambridge University Press, Cambrige, 1991.

(Lave, 1997):

Lave, J.: The culture of acquisition and the practice of understanding. In:

Kirshner, D.; Whitson, J.A. (eds.): Situated cognition: Social, semiotic, and

psychological perspectives, Mahwah, NJ, 1997, p.63-82.

(Lea and Spears, 1991):

Lea, M.; Spears, R.: Computer-mediated communication, deindividuation and

group decision making. International Journal of Man-Machine Studies, 34

(1991), p. 283-301.

(Lehner, 2000):

Lehner, F.: Organisational Memory – Konzepte und Systeme für das organisa-

torische Lernen und das Wissensmanagement (in German). Hanser-Verlag

München, Wien, 2000.

(Leihmeister et al., 2004):

Leimeister, J.M.; Sidiras, P.; Krcmar, H.: Success factors of virtual communi-

ties from the perspective of members and operators: An empirical study. Pro-

ceedings of 37th IEEE Hawaii International Conference of Systems Sciences

HICCS37, Hawaii, USA, 2004.

(Lesniak et al., 1977):

Lesniak, R.;Yates, M.; Goldhaber, G. M.; Richards; W.: NEGOPY and

NETPLOT - Program Characteristics. Connections, (1)1977, p. 26-29.

(Lesser and Cothrel, 2001):

Lesser, E.; Cothrel, J.P.: Fast Friends: Virtuality and Social Capital. Knowl-

edge Directions, Spring/Summer 2001.

250 M.Trier

(Lesser and Prusak, 1999):

Lesser, E.; Prusak, L.: Communities of Practice, Social Capital and Organiza-

tional Knowledge. White Paper, IBM Institute for Knowledge Management,

August 1999.

(Lesser and Storck, 2001):

Lesser, E. L.; Storck, J.: Communities of practice and organizational perform-

ance. IBM SYSTEMS JOURNAL, 40(2001)4, p.831-841.

(Lichtblau, 1998):

Lichtblau, K.: Beschäftigungsentwicklung, Strukturwandel und Qualifikati-

onsprofil des Humankapitals (in German), iw-trends (25) 1998, p. 15-31.

http:// www.ifo.de/pls/portal30 /docs/

FOLDER/IFO_PORTAL/IFO_INSTITUT /WIR_UEBER

_UNS_X_ABOUT_IFO/BEREICHE/STRUKTURWANDEL

_UND_BRANCHEN_ SB_/ PROJEKTE_SB/MASCHINENBAU-STUDIE

.PDF, Accessed 2004-03-01.

(Locke, 1690):

Locke, J.: An Essay concerning Human Understanding. Complete and

unabrigded edition, collated and annotated by A.C. Fraser, New York, 1959.

(Luhmann, 1984):

Luhmann, N.: Soziale Systeme – Grundriss einer allgemeinen Theorie. Frank-

furt am Main, 1984.

(Malone et al., 1987):

Malone, T.; Yates, J.; Benjamin, R.: Electronic markets and electronic hierar-

chies: Effects of information technology on market structure and corporate

strategies. Communications of the ACM, 30(1987)6, p. 484-497.

(Mantovani, 1996):

Mantovani, G.: New communication environments: from everyday to virtual.

London, 1996.

(March and Olson, 1975):

March, J.G.; Olson, J.P.: The Uncertainty of the Past: Organizational Learning

under Ambiguity. In: European Journal on Political Research, 3(1975), p.147-

171.

(March and Simon, 1958):

March, J.; Simon, H. Organizations. New York, 1958.

(Marsden and Campbell, 1984):

Marsden, P.; Campbell, K. E.: Measuring tie strength. Social Forces, 63(1984),

p. 482-501.

Visualization of virtual Knowledge Communities 251

(Mayo, 1949):

Mayo, E. : The social problems of an industrial civilization. New York, 1949.

(Mayo, 1960):

Mayo, E.: The human problems of an industrial civilization. New York, 1960.

(McDermott, 1999):

McDermott, R.: Nurturing Three Dimensional Communities of Practice: How

to get the most out of human networks. In: Knowledge Management Review,

Fall 1999.

(McDermott, 2000):

McDermott, R.: Knowing in Community: 10 Critical Success Factors in Build-

ing Communities of Practice. IHRIM Journal, March 2000.

(McLureWasko & Faraj, 2000):

McLure-Wasko, M.; Faraj, S.: It is what one does: why people participate and

help others in electronic communities of practice. Journal of Strategic Informa-

tion Systems, 2-3(2000)9, p. 155-173.

(McWilliam, 2000):

McWilliam, G.: Building Strong Brands Through Online Communities. Sloan

Management Review, Spring 2000.

(Mead, 1934):

Mead, G.H.: Mind, Self and Society from the Standpoint of a Social Behav-

iourist. Chicago 1934. http://spartan.ac.brocku.ca/~lward/mead/pubs

/Mead_1909a.html, Accessed: 2004-06-06.

(Miller et al., 2002):

Miller, D. R.; Fontaine, A.; Muller, M. J.: Understanding the Benefits and

Costs of Communities of Practice, 2002.

http://domino.watson.ibm.com/cambridge/ re-

search.nsf/2b4f81291401771785256976004a8d13/4226131748e

15f0785256ba600536214/$FILE/TR2002.01.pdf, Accessed: 2002-11-03.

(Milton, 2003):

Milton, N.: Information on Knowledge Acquisition. 2003.

http://www.epistemics.co.uk/ Notes/37-0-0.htm, Accessed 2004-07-01.

(Mintzberg, 1992):

Mintzberg, H.: Die Mintzberg-Struktur. Organisationen effektiver gestalten (in

German). Landsberg/Lech, 1992.

(Moreno, 1932):

Moreno, J. L.: Application of the Group Method to Classification. National

Committee on Prisons and Prison Labor, New York, 1932.

252 M.Trier

(Moreno, 1934):

Moreno, J. L.: Who Shall Survive? Washington, DC: Nervous and Mental Dis-

ease Publishing Company, 1934.

(Nahapiet and Ghoshal, 1998):

Nahapiet, J.; Ghoshal, S.: Social Capital, Intellectual Capital and the organiza-

tional Advantage. Academy of Management Review, April 1998, p. 243.

(Nassehi, 2000):

Nassehi, A.: Was wissen wir über das Wissen (in German). Symposium

Szenarien der Wissensgesellschaft. München, October 2000.

(Nickols, 2000):

Nickols, F.: Communities of Practice - Two Types of CoPs.

http://home.att.net/~discon/KM/CoPTypes.htm, Accessed 2003-02-07.

(Nonaka and Takeuchi, 1995):

Nonaka, I.; Takeuchi, H.: The Knowledge-Creating Company. How Japanese

Companies Foster Creativity and Innovation for Competitive Advantage. Lon-

don et al., 1995.

(Nonaka and Takeuchi, 1997):

Nonaka, I.; Takeuchi, H.: Die Organisation des Wissens: wie japanische Un-

ternehmen eine brachliegende Ressource nutzbar machen (in German).. Frank-

furt am Main et al., 1997.

(Nonaka and Teece, 2001):

Nonaka, I.; Teece, D. J.: Managing industrial knowledge - Creation, transfer

and utilization. London et al., 2001, p. 344.

(Nonaka, 1991):

Nonaka, I.: The Knowledge-Creating Company. In: Harvard Business Review,

November 1991, p.96-104.

(North and Papp, 1999):

North, K.; Papp, A.: Wissensmanagement implementieren (in German).

REFA-Nachrichten 51, Nr.6, p.20-27.

(North et al., 2000):

North, K.; Romhardt, K.; Probst, G.: Wissensgemeinschaften (in German). IO-

Management, 69(2000)7/8, 2000, p. 52-62.

(North, 1999):

North, K.: Wissensorientierte Unternehmensführung (in German). Wiesbaden,

1999.

Visualization of virtual Knowledge Communities 253

(OECD, 2000):

Organisation for Economic Cooperation and Development: International Di-

rect Investment Statistics Yearbook, 2000, Paris.

http://www.oecd.org/dataoecd/38 /28/1938009.pdf, Accessed: 2004-03-05.

(OECD, 2001b):

Organisation for Economic Cooperation and Development: New Patterns of

Industrial Globalisation: Cross-border Mergers and Acquisitions and Strategic

Alliances, Paris, August 2001. http://www.oecd.org/dataoecd/38

/28/1938009.pdf, Accessed: 2004-03-05.

(OECD, 2002a):

Organisation for Economic Cooperation and Development: OECD Business

and Industry Policy Forum Reports on Proceedings, February 2002.

http://www.oecd.org/dataoecd/38/28 /1938009.pdf, Accessed: 03-06-2005.

(OECD, 2002b):

Organisation for Economic Cooperation and Development: OECD Economic

Outlook, No. 71, Paris, 2002. http://www.oecd.org/dataoecd/38

/28/1938009.pdf, Accessed: 2004-03-05.

(Oliver et al., 1998):

Oliver, R.; Omari, A.; Herrington, J.: Exploring student interactions in collabo-

rative World Wide Web computer-based learning environments. Journal of

Educational Multimedia and Hypermedia, 2/1998, p. 263–287.

(Ortmann, 1990):

Ortmann, G.: Computer und Macht in Organisationen (in German).Opladen,

1990.

(Ouchi, 1977):

Ouchi, W.G.: The relationship between Organizational Structure and Organ-

izational Control. In: Administrative Science Quarterly, 22(1977), p.95-113.

(Palmer and Rock, 1994):

Palmer, S. E.; Rock, I.: Rethinking perceptional organization: The Role of Uni-

form connectedness. Psychonomic Bulletin and Review 1(1994)1, p.29-55.

(Parsons, 1937):

Parsons, T.: The structure of social action: a study in social theory with special

reference to a group of recent European writers. New York, 1937.

(Parsons, 1951):

Parsons, T.: The Social System. Glencoe, 1951.

(Participate, 2001):

Participate: Driving Sales Force Effectiveness Through Knowledge Sharing.

White Paper, 2001.

254 M.Trier

(Pautzke, 1989):

Pautzke, G.: Die Evolution der organisatorischen Wissensbasis: Bausteine zu

einer Theorie des organisatorischen Lernens (in German). München, 1989.

(Pawlowski, 1994):

Pawlowski, P.: Wissensmanagement in der lernenden Organisation (in Ger-

man). Habilitation. Paderborn, 1994.

(Pennings and Lee, 1998):

Pennings, J. M.; Lee, K. Social Capital of Organization: Conceptualization,

Level of Analysis and Performance Implications. In: Gabbay, S.; Leenders, R.

(eds.): Corporate Social Capital, New York, 1998.

(Penrose, 1958):

Penrose E.P.: The Theory of the Growth of the Firm. Oxford, 1959.

(Perillieux, 1991)

Perillieux, R.: Strategisches Timing von F&E und Markteintritt bei innovati-

ven Produkten (in German). In: Booz-Allen & Hamilton (eds.): Integriertes

Technologie- und Innovationsmanagement (in German), Berlin 1991, p. 21-48.

cit in. http://www.ifbf.unizh.ch/orga/downloads/vorlesungen/TIM.pdf, Acces-

sed 2004-03-01.

(Pfeffer and Salancik, 1978):

Pfeffer, J.; Salancik, G.R.: The external control of Organizations. New York,

1978.

(Plato, 1977):

Plato: Phaidon, 65e-66a (In German). In: Plato: Sämtliche Werke (in German),

Band 3, Hamburg, 1977.

(Polanyi, 1985):

Polanyi, M.: Implizites Wissen (in German). Frankfurt am Main, 1985.

(Por, 1997):

Por, G.: Designing Knowledge Ecosystems for Communities of Practice. Pres-

entation at the ICM conference on Knowledge Management, Los Angeles,

1997.

(Powell, 1990):

Powell, W. W.: Neither market nor hierarchy: Network forms of organization.

Research in Organizational Behavior, 12(1990).

(Powell, 1996):

Powell, W.: Weder Markt noch Hierarchie: Netzwerkartige Organisationsfor-

men (in German). In: Kerris, P.; Schneider, V. (eds.): Organisation und Netz-

werk. Institutionelle Steuerung in Wirtschaft und Politik. Frankfurt am Main,

New York, 1996.

Visualization of virtual Knowledge Communities 255

(Probst and Buechel, 1994):

Probst, G.J.B.; Buechel, B.: Organisationales Lernen, Wettbewerbsvorteil der

Zukunft (in German). Wiesbaden, 1994.

(Probst et al., 1997):

Probst, G.; Romhardt, K.; Raub, S.: Wissen managen – Wie Unternehmen ihre

wertvollste Ressource optimal nutzen. Frankfurt, 1997.

(Proctor, 1953):

Proctor, C.: Informal social systems. In: Loomis, C. P.; Moralis, J. O.; Clifford,

R. A.; Leonard, O. E. (eds.): Turrialba. Glencoe, 1953, p. 73-88.

(Rao and Goldmann-Segall, 1995):

Rao, V. S.; Goldmann-Segall, R.: Capturing Stories in Organizational Mem-

ory. In: Proceedings of the 28th Hawaii International Conference on Systems

Sciences 1995, Los Alamitos, 1995, p.333-341.

(Rehaeuser and Krcmar, 1996):

Rehaeuser, J.; Krcmar, H.: Wissensmanagement im Unternehmen. In:

Schreyoegg, G.; Conrad, P. (eds.): Managementforschung, Volume 6. Berlin,

New York, 1996, p.263-304.

(Reinhardt and Beyer , 1997):

Reinhardt, R.; Beyer, J.: Wissensmanagement. Ein neuer Ansatz für das Ma-

nagement und eben kein Wein in alten Schläuchen (in German). In: Wis-

sensmanagement. Sonderheft Süddeutsche Zeitung, 1997, p.35-37.

(Reinhardt and Pawlowski, 1997):

Reinhardt, R.; Pawlowski, P.: Wissensmanagement: Ein integrativer Ansatz

zur Gestaltung organisationaler Lernprozesse (in German). In: Handbuch

Lernende Organisation. Wiesbaden, 1997.

(Reinmann-Rothmeier and Mandl, 1997):

Reinmann-Rothmeier, G.; Mandl, H.: Wissensmanagement: Phänomene - Ana-

lyse – Forschung - Bildung (in German). Research Report Nr. 83, LMU Mu-

nich, October 1997.

(Reinmann-Rothmeier, 2000):

Reinmann-Rothmeier, G.: Communities und Wissensmanagement. Wenn hohe

Erwartungen und wenig Wissen zusammentreffen (in German). Research Re-

port Nr. 129, LMU Munich, December 2000.

(Resnick, 2002):

Resnick, P.: Beyond Bowling Together: SocioTechnical Capital. In: "HCI in

the New Millenium", edited by John M. Carroll. Addison-Wesley. 2002, p.

247-272. http://www.si.umich.edu/~presnick/papers/stk/ResnickSTK.pdf, Ac-

cessed: 2004-03-10.

256 M.Trier

(Rheingold, 1993):

Rheingold, H.: The Virtual Community - Homesteading On The Electronic

Frontier. New York, 1993.

(Riva and Galimberti, 1997):

Riva, G.; Galimberti, C.: The psychology of cyberspace: a sociocognitive

framework to computer mediated communication. New Ideas in Psychology,

15(1997)2, p.141-158.

(Roberts, 1998):

Roberts, T. L.: Are Newsgroup Virtual Communities? Proceedings of the

ACM CHI, April 1998, p.360-367.

(Rocco, 1998):

Rocco, E.: Trust Breaks Down in Electronic Contexts But Can Be Repaired by

Some Initial Face-to-Face Contact. Proceedings of CHI-98, New York, 1998.

(Roethlisberger and Dickson, 1947):

Roethlisberger, F. J.; Dickson, W. J.: Management and the worker: An account

of a research program conducted by the Western Electric Company, Haw-

thorne Works, Chicago, Cambrigde, 1947.

(Roschelle, 1996):

Roschelle, J.: Designing for Cognitive Communication: Epistemic Fidelity or

Mediating Collaborative Inquiry. In: Day, D.; Kovacs, D. (eds.): Computers

Communication and Mental Models, Taylor and Francis, 1996.

(Russell, 1954):

Russell, B.: Philosophie des Abendlandes: ihr Zusammenhang mit der politi-

schen und sozialen Entwicklung (in German). Darmstadt, 1954.

(Sachs, 1995):

Sachs, P.: Transforming Work: Collaboration, Learning, and Design. Commu-

nications of the ACM, 38(1995)9, p. 36-44.

(Sackmann, 1992):

Sackmann, S.: Culture and Subcultures: An Analysis of Organizational

Knowledge. In: Administrative Science Quarterly, 32(1992), p.140-161.

(Sartre, 1980):

Sartre, J.-P.: Das Sein und das Nichts (in German). Reinbek, 1980.

(Schaefers, 1980):

Schäfers, B.: Einführung in die Gruppensoziologie. Heidelberg, 1980.

(Schein, 1980):

Schein, E. H.: Organizational Psychology. Englewood Cliffs, N.J., 1980.

Visualization of virtual Knowledge Communities 257

(Schneider, 1996):

Schneider, U.: Wissensmanagement. Die Aktivierung des intellektuellen Kapi-

tals (in German). Frankfurt, 1996.

(Schoen, 2000):

Schoen, S.: Gestaltung und Unterstützung von Communities of Practice (in

German). Dissertation, Technical University Munich, 2000.

(Schreiber et al., 2000):

Schreiber, G.; Akkermans, H.; Anjewierden, A.; de Hoog, R.; Shadbold, N.;

van der Velde, W.;Wielinda, B.: Knowledge Engineering and Management,

The CommonKADS Methodology. The MIT Press, Cambrigde, 2000.

(Schueppel, 1996):

Schueppel, J.: Wissensmanagement. Organisatorisches Lernen im Spannungs-

feld von Wissens- und Lernbarrieren. Wiesbaden, 1996.

(Schulz von Thun, 1981):

Schulz von Thun, F.: Miteinander reden. Differentielle Psychologie der Kom-

munikation (in German). Reinbeck, 1981.

(Schwaniger, 1995):

Schwaniger, M.: Systemtheorie. Eine Einführung für Führungskräfte, Wirt-

schaftler und Sozialwissenschaftler (in German). In: In: Institut für Betrieb-

swirtschaft – Disskussionsbeiträge, Universität St. Gallen, 1995.

(Scott, 1986):

Scott, W. R.:Grundlagen der Organisationstheorie (in German). Frankfurt am

Main, New York, 1986.

(Scott, 1991):

Scott, J.: Social Network Analysis: A Handbook. SAGE Publication Ltd. Lon-

don, 1991.

(Segler, 1985):

Segler, T.: Die Evolution von Organisationen. Ein evolutionstheoretischer An-

satz zur Erklärung der Entstehung und des Wandels von Organisationsformen

(in German). Frankfurt am Main et al., 1985.

(Seidel and Lehner, 1999):

Seidel, S.; Lehner, F.: Wissensmanagement: Begriffsauffassung und Umset-

zung in der Praxis (in German). Research Report. University of Regensburg,

December 1999.

(Senge, 1990):

Senge, P.: The fifth Discipline. New York, 1990.

258 M.Trier

(Seufert et al., 1999):

Seufert, A.; von Krogh, G.; Bach, A.:Towards knowledge networking. Journal

of Knowledge Management, (3)1999, p.180-190.

(Sfard, 1998):

Sfard, A.: On two metaphors for learning and the dangers of choosing just one.

Educational Researcher, 27(1998), p. 4-13.

(Shannon and Weaver, 1949):

Shannon C. E.; Weaver W.: The Mathematical Theory of Communication. Ur-

bana, Ill., Illinois UP, 1949.

(Shell, 2001):

Shell International Exploration and Production: Stories from the Edge: Manag-

ing Knowledge through New Ways of Working within Shell’s Exploration and

Production Business. White Paper, 2001.

(Slocum, 1983):

Slocum, T. S.: Predicting Visual Clusters on graduated Circle Maps. American

Cartographer, 10(1983)1, p. 59-72.

(Smith and Fiore, 2001):

Smith, M. A.; Fiore, A. T.: Visualization components for persistent conversa-

tions. Proceedings of the SIGCHI conference on Human factors in computing

systems, 2001.

(Smith, 1767):

Smith, A.: An inquiry into the nature and causes of the wealth of nations.

Strahan and Cadell, London. http://www.adamsmith.org/smith/won-intro.htm,

Accessed 2004-10-01.

(Snis, 2001):

Snis, U. L.: Challenges towards knowledge communities in engineering do-

mains: Adding culture to knowledge management. Proceedings of IRIS24

Ulvik, Norway, 2001.

(Snyder, 1997):

Snyder, W.M.: Communities of Practice: Combining Organizational Learning

and Strategy Insights to Create a Bridge to the 21st Century, Working Paper,

Social Capital Group. Cambridge, Massachussets, 1997.

(Snyder, 1999):

Snyder, W.M.: Organization and World Design: The Gaia’s Hypotheses.

Working Paper. Social Capital Group. Cambridge, Massachussets, 1999.

(Sodian, 1986):

Sodian, B.: Wissen durch Denken (in German). Aschendorf, Muenster, 1986.

Visualization of virtual Knowledge Communities 259

(Spears and Lea, 1994)

Spears, R.; Lea, M.: Panacea or Panopticon? The Hidden Power in Computer-

Mediated Communication. Communication Research, 21(1994), p.427-453.

(Spiegel, 2001):

Spiegel, D.: Coterie: A Visualization of the Conversational Dynamics within

IRC. Thesis, MIT Media Labs: Cambridge, USA, 2001.

(Sproull and Kiesler, 1986):

Sproull, L.; Kiesler, S.: Reducing social context cues: Electronic mail in organ-

izational communication. Organizational Behavior and Human Decision Proc-

esses, 37(1986), p.157-187.

(Sproull and Kiesler, 1991):

Sproull, L.; Kiesler, S.: Connections: New ways of working in the networked

organizations. Cambridge, MA, 1991.

(Squire, 1992):

Squire, L.R.: Declarative and nondeclarative memory: Multiple brain systems

supporting learning and memory. In: Journal of Cognitive Neuroscience,

4(1992), p. 232-243.

(Staehle, 1991):

Staehle, W.: Management: Eine verhaltenswissenschaftliche Perspektive,

München, 1991.

(Stamp, 1997):

Stamp D.: Community of Practice – Learning is Social. Training is Irrelevant?

Training, February 1997, p.35-42.

(Stamps and Lipnack, 2000):

Stamps, J.; Lipnack, J.: A Systems Science of Networked Organisations. Pro-

ceedings of the World Congress on Systems Sciences ISSS 2000.

www.virtualteams.com/library/whpapers/ ISSS_2000.pdf, Accessed:

01.10.2004.

(Steward and Brown, 1996):

Steward, T. A.; Brown, V.: The invisible key to success. In: Fortune, August

5(1996), p. 1-4.

(Storck and Hill, 2000):

Storck, J.; Hill, P.: Knowledge Diffusion Through ‘Strategic Communities’.

Sloan Management Review 41(2000)2.

(Strube et al., 1996):

Strube. G.; Becker, B.; Freksa, C.; Hahn, U.; Opwis, K.; Palm, G.: (eds.): Wör-

terbuch der Kognitionswissenschaft (WbKW) (in German). Stuttgart, 1996.

260 M.Trier

(Swan et al., 2000):

Swan, J.; Newell, S. ; Robertson, M.: Limits of IT-Driven Knowledge Man-

agement Initiatives for Interactive Innovation Processes: Towards a Commu-

nity-Based Approach. In Proceedings of the 33rd Hawaii International Confer-

ence on System Sciences, Hawaii, 2000.

(Swan, 2001):

Swan, J.: Knowledge Management in Action: Integrating Knowledge across

Communities, In Proceedings of the 34rd Hawaii International Conference on

System Sciences, Hawaii, 2001.

(Taylor, 1911):

Taylor, F.W.: Principles of Scientific Management. New York, 1911. Also

published in 1983.

(Teubner, 1992):

Teubner, G.: Die vielköpfige Hydra: Netzwerke als kollektive Akteure höherer

Ordnung (in German). In: Krohn, W. / Küppers, G. (eds.): Emergenz. Die Ent-

stehung von Ordnung, Organisation und Bedeutung. Frankfurt a.M., 1992, p.

189–216.

(Thomas et al., 2001):

Thomas, J. C.; Kellogg, W. A.; Erickson, T.: The knowledge management puz-

zle: Human and social factors in knowledge management. IBM Systems Jour-

nal, 40(2001)4.

(Tichy et al., 1979):

Tichy, N. M.; Tushman, M. L.; Fombrun, C.: Social Network Analysis for Or-

ganizations. Academy of Management Review, 4(1979)4, p.507-519.

(Trier and Mueller, 2004):

Towards a systematical approach for Capturing Knowledge-intensive Business

Processes. In: Karagiannis, D.; Reimer, U.(eds.):Lecture Notes in Artificial In-

telligence Volume 3336, Subseries of Lecture Notes in Computer Science,

Berlin et al., 2004, p.239-250.

(Trier, 2002):

Trier, M.: Communities - Management und Unterstützung (in German). In:

Gronau, N. (Ed.): Wissensmanagement - Strategien, Prozesse, Communities.

Aachen, 2002, p.149-159.

(Trier, 2002):

Trier, M.: Integrating Communities with E-Learning Platforms to improve on-

the-job learning. Proceedings OKLC, Barcelona, 2002.

(Trier, 2003):

Trier, M.: Anforderungen an die IT-Unterstützung für Communities of Practice

(in German). Industrie Management, 3(2003), p.49-53.

Visualization of virtual Knowledge Communities 261

(Trier, 2003):

Trier, M.: IT-gestütztes Management virtueller Communities of Practice (in

German). Extended Abstract of DoIT-Kongress Karlsruhe 2003.

http://www.integrata-stiftung.de/ files/p14_trier.pdf, Accessed 2003-12-12.

(Trier, 2005):

Trier, M.: IT-supported Visualization of Knowledge Community Structures.,

Proceedings of 38th IEEE Hawaii International Conference of Systems Sci-

ences, Hawaii, USA, 2005.

(Trommsdorff, 1990):

Tromsdorf, V.: Erfolgsfaktorenforschung, Produktinnovation und Schnittstelle

Marketing-F&E (in German). Diskussionspapier/Technische Universität Ber-

lin, Wirtschaftswissenschaftliche Dokumentation. Technische Universität Ber-

lin, 1990.

(Tsai and Ghoshal, 1998):

Tsai, W.; Ghoshal, S.: Social Capital and Value Creation: The Role Intrafirm

Networks. Academy of Management Review, August 1998, p. 464.

(Tuckman, 1965):

Tuckman, B. W.: Developmental Sequence in Small Groups. In: Psychological

Bulletin, 63(1965), p.384-399.

(Tyler et al., 2003):

Tyler J.; Wikinson D.; Huberman B.A.: E-Mail Spectroscopy. http://www.

Hpl.hp.com/shl/papers/email/index.html. Accessed: 2003-05-13.

(Ulrich et al., 1976):

Ulrich, H.; Krieg, W.; Malik, F.: Zum Praxisbezug einer systemorientierten

Betriebswirtschaftslehre (in German). In: Ulrich, H. (Ed.): Zum Praxisbezug

einer systemorientierten Betriebswirtschaftslehre in wissenschaftstheoretischer

Sicht. Bern, Stuttgart, 1976, p.135-151.

(UNIDO, 2002):

UNIDO United Nations Industrial Development Organization. Industrial De-

velopment Report – Competing through innovation and learning. 2002/2003.

http://www.unido.org/userfiles/hartmany/12IDR_full_report.pdf, Accessed:

2004-03-01.

(VDI, 1995):

Verein Deutscher Ingenieure: VDI Richtlinie 5007 – Wissensbanken in der

Anwendung (in German), 1995.

(Vestal, 2003):

Vestal, W.: Ten traits for a successful Community of Practice, KM Review

5(2003)6.

262 M.Trier

(Vico, 1710):

Vico, G.-B.: De antiquissima Italorum sapientia. Napels, 1858. Cit. in. Glasers-

feld, 1996, cit.op.

(Vygotsky, 1930):

Vygotsky, L.: Mind in Society. Cambridge, MA, 1930.

(Walsh and Ungson, 1991):

Walsh, J. P.; Ungson, G. R.: Organizational Memory. In: Academy of Man-

agement Review, 16(1991)1, p.57-91.

(Walther and Burgoon, 1992):

Walther, J. B.; Burgoon, J. K.: Relational Communication in Computer-

Mediated In-teraction. Human Communication Research, 19(1992), p.50-81.

(Ware, 2000):

Ware, C.: Information Visualisation: Using Vision to Think. San Diego, 2000.

(Wargitsch, 1998):

Wargitsch, C.: Ein Beitrag zur Integration von Workflow- und Wissensmana-

gement unter besonderer Berücksichtigung komplexer Geschäftsprozesse. Dis-

sertation. Erlangen-Nuernberg, 1998.

(Wasserman and Faust, 1994):

Wasserman, S.; Faust, K.: Social Network Analysis: Methods and Applica-

tions. Cambridge University Press: Cambridge, 1994.

(Watson, 1996):

Watson, R.T.: Data Management. An Organizational Perspective, New York et

al., 1996.

(Watzlawick et al., 1985):

Watzlawick, P.; Beavin, J. H.; Jackson, D. D.: Menschliche Kommunikation

(in German). 7th Edition, Bern, 1985.

(Webber, 1999):

Webber, A. M.: Learning for a change. Fast Company, 24(1999) May, p.178.

http://www.fastcompany.com/magazine/24/senge.html, Accessed: 2004-12-12.

(Weber, 1972):

Weber, M.: Wirtschaft und Gesellschaft. Grundriß einer verstehenden Soziolo-

gie (in German). 5th Edition. Tübingen, 1972.

(Wegner, 1987):

Wegner, D. M.: Transactive Memory: A Contemporary Analysis of the Group

Mind. In: Mullen, B.; Goethals, G. R. (eds.) Theories of Group Behaviour,

New York et al., 1987, p.185-208.

Visualization of virtual Knowledge Communities 263

(Weick and Roberts, 1993):

Weick, K. E.; Roberts, K. H.: Collective Mind in Organizational Hedful Inter-

relating on Flight Decks. In: Administrative Science Quarterly 38(1993)3,

p.357-381.

(Weick, 1991):

Weick, K. E.: The Nontraditional Quality of Organizational Learning. In: Or-

ganization Science, 2(1991)1, p.116-124.

(Wellman and Wortley, 1990):

Wellman, B.; Wortley, S.: Different Strokes from Different Folks: Community

Ties and Social Support. American Journal of Sociology. 96(1990), p.558-88.

(Wellman, 1997):

Wellman, B.: An Electric Group is Virtually a Social Network. Culture of the

Internet, p. 179-205.

(Wellmann and Potter, 1997):

Wellman, B.; Potter, S.: The elements of personal community. In: Wellman, B.

(ed.): Networks in the global village. Norwood, NJ, 1997.

(Wellmann et al., 1996):

Wellman, B.; Salaff, J.; Dimitrova, D.; Garton, L.; Gulia, M.; Haythorn-

thwaite, C.: Computer networks as social networks: Collaborative work, tele-

work, and virtual community. Annual Review of Sociology, 22(1996), p. 213-

238.

(Wenger and Snyder, 2000):

Wenger, E.C.; Snyder, W.M.: Communities of Practice: The Organizational

Frontier. In: Harvard Business Review, January 2000, p.139-145.

(Wenger et al., 2002):

Wenger, E.; McDermott, R.; Snyder, W.: Cultivating Communities of Practice:

A Guide to Managing Knowledge. Boston, 2002.

(Wenger, 1998):

Wenger, E.: Communities of Practice: Learning as a Social System, Systems

Thinker, 9(5)1998.

(Wenger, 1998a):

Wenger, E.C.: Communities of Practice: Learning, Meaning, and Identity,

Cambridge, UK, 1998.

(Weyer et al., 1997):

Weyer, J. (ed.): Konzepte und Methoden der sozialwissenschaftlichen Netz-

werkforschung (in German). München, Wien, 1997.

264 M.Trier

(Wiegand, 1996):

Wiegand, M.: Prozesse organisationalen Lernens (in German). Wiesbaden,

1996

(Wiener, 1948):

Wiener, N.: Cybernetics or control and communication in the animal and the

machine. Cambridge, 1948

(Wiese, 1924):

Wiese von, L.: System der Allgemeinen Soziologie als Lehre von den sozialen

Gebilden der Menschen (Beziehungslehre) (in German). Berlin, 1924.

(Willke, 1995):

Willke, H.: Systemtheorie III: Steuerungstheorie: Grundzüge einer Theorie der

Steuerung komplexer Sozialsysteme, Stuttgart et al.,1995.

(Willke, 1996):

Willke, H.: Dimensionen des Wissensmanagements – Zum Zusammenhang

von gesellschaftlicher und organisationaler Wissensbasierung (in German). In:

Schreyoegg, G.; Conrad, P. (Eds.): Managementforschung, Volume 6. Berlin,

New York, 1996, p.263-304.

(Willke, 1998):

Willke, H.: Systemisches Wissensmanagement, Stuttgart, 1998.

(Windeler, 2001):

Windeler, A.:Unternehmensnetzwerke. Konstitution und Strukturation (in Ger-

man). Wiesbaden, 2001.

(Winograd and Flores, 1987):

Winograd, T.; Flores, C. F.:Understanding Computers and Cognition. A new

Foundation for Design. Reading, Massachussets, 1987.

(Wyssusek et al.; 2001):

Wyssusek, B.; Schwartz, M.; Kremberg, B.: Knowledge Management – A So-

ciopragmatic Approach. In: Proceedings of the 2nd European Conference on

Knowledge Management, Slovene, 2001, p.767-776.

(Wyssusek, 2004):

Wyssusek, B.: Methodologische Aspekte der Organisationsmodellierung in der

Wirtschaftsinformatik, Doctoral Dissertation, TU Berlin, 2004.

(Xiong and Donath, 1999):

Xiong, R.; Donath, J.: PeopleGarden: Creating Data Portraits for Users. Pro-

ceedings 12th Annual ACM Symposium on User Interface Software and Tech-

nology. New York, 1999, p. 37-44.

Visualization of virtual Knowledge Communities 265

(ZEW, 2003):

Zentrum für Europäische Wirtschaftsforschung: ZEW Branchenreport (in

German). Zentrum für Europäische Wirtschaftsforschung, Mannheim. Volume

10, Nr. 8, Mai 2003. ftp://ftp.zew.de/pub/zew-docs/brarep_inno/Brarep-

Maschinenbau. pdf, Accessed: 2004-03-05.

(Zucker, 1986):

Zucker, L.G.: Production of trust: Institutional sources of economic structure,

1840-1920. Research in Organizational Behavior, 8/1986, p.53-111.

266 M.Trier

12 Appendices

12.1 CoPs in corporate KM Approaches – Overview

Overview about recent corporate approaches to CoPs and KM:

(cf.: http://www.providersedge.com/docs/km_articles/Measuring_KM.pdf)

Organization Target Value

Proposition Approach Results

ChevronTexaco Reduce operating costs,

improve operational

excellence, improve

safety

CoPs, facilitate transfer of

best practices, People Finder Two billion dollar reduction in

annual operating costs (1991 v.

1998); $670 million came from

refining best practices.

Total investment of more than

$2 million (total figure unknown)

Dow Chemical Provide faster access to

information, improve

information

management, improve

sales leads

Content management,

communities of practice Increased number of sales leads

—Increase in new product sales

Improved customer satisfaction

scores

CM investment of over $3

million for start up, $8 million

annually.

GE Plastics Decrease customer

service costs Customer portal, customer

knowledge repository Number of test chips created

decreased from 4.2 to 2.7

Average reduction of 4.5 hours per

color match

Savings of $2.25 million per year

Total investment unknown

Shell Create a single, global

company

Reduce cycle time

“Too fast to follow”

Global Networks (CoPs)

New ways of working

Letting the new guys into “Old

Boy” networks

Transfer of best practices

$200 million/yr cost savings

Reduced number of wells

Increased facility uptime

Reduced design and planning

errors

Total investment of

approximately $4 million

BP Know-how:

A brand attribute;

ability to innovate and

execute faster and

smarter than competitors

Networks, Peer Assist, AARs,

Retrospects, Technology VP

support, Operations Value

Process

$260 million cost savings

Drilling cost reduction for

Schiehallion

(West of Shetlands)

Refinery turnarounds

Retail site construction , Total

investment unknown

Schlumberger Knowledge in the hands

of employees and

customers

CoPs, InTouch KM system,

intranet, extranet, content

management

$200 million cost savings

95 percent reduction in time to

resolve tech queries

75 percent reduction in updating

modifications

Total investment of

approximately $20 million

Cap Gemini Ernst &

Young Faster revenue growth,

lower costs CoPs, central KM managers,

content management Ten-fold increase in revenue with

only five-fold increase in

employees

IBM Global Services Revenue growth,

industry leadership COPs, knowledge managers,

Intellectual Capital

Management System

400 percent increase in service

revenue

Time savings of $24 million in

1997

Approximately $750K to start

up, $750K annually to maintain

Best Buy Bring creative new

solutions to market

faster, shorten the

learning curve, lower

costs

Portal (RetailZone),

Employee Toolkit,

Communities of Practice

(Retail and Services)

1.5 percent increase in gross

margin

Sold 4.2 units/store/day more in

pilot stores

3 percent drop in damage claims

Paper reduction savings of

$250K/year

Total investment of

approximately $3.5 million

Data gleaned from APQC’s “Successfully Implementing KM” (1999), “Managing Content and Knowledge” (2001), and “Retaining

Valuable Knowledge” (2002) benchmarking studies

Organization Target Value

Proposition Approach Results

ChevronTexaco Reduce operating costs,

improve operational

excellence, improve

safety

CoPs, facilitate transfer of

best practices, People Finder Two billion dollar reduction in

annual operating costs (1991 v.

1998); $670 million came from

refining best practices.

Total investment of more than

$2 million (total figure unknown)

Dow Chemical Provide faster access to

information, improve

information

management, improve

sales leads

Content management,

communities of practice Increased number of sales leads

—Increase in new product sales

Improved customer satisfaction

scores

CM investment of over $3

million for start up, $8 million

annually.

GE Plastics Decrease customer

service costs Customer portal, customer

knowledge repository Number of test chips created

decreased from 4.2 to 2.7

Average reduction of 4.5 hours per

color match

Savings of $2.25 million per year

Total investment unknown

Shell Create a single, global

company

Reduce cycle time

“Too fast to follow”

Global Networks (CoPs)

New ways of working

Letting the new guys into “Old

Boy” networks

Transfer of best practices

$200 million/yr cost savings

Reduced number of wells

Increased facility uptime

Reduced design and planning

errors

Total investment of

approximately $4 million

BP Know-how:

A brand attribute;

ability to innovate and

execute faster and

smarter than competitors

Networks, Peer Assist, AARs,

Retrospects, Technology VP

support, Operations Value

Process

$260 million cost savings

Drilling cost reduction for

Schiehallion

(West of Shetlands)

Refinery turnarounds

Retail site construction , Total

investment unknown

Schlumberger Knowledge in the hands

of employees and

customers

CoPs, InTouch KM system,

intranet, extranet, content

management

$200 million cost savings

95 percent reduction in time to

resolve tech queries

75 percent reduction in updating

modifications

Total investment of

approximately $20 million

Cap Gemini Ernst &

Young Faster revenue growth,

lower costs CoPs, central KM managers,

content management Ten-fold increase in revenue with

only five-fold increase in

employees

IBM Global Services Revenue growth,

industry leadership COPs, knowledge managers,

Intellectual Capital

Management System

400 percent increase in service

revenue

Time savings of $24 million in

1997

Approximately $750K to start

up, $750K annually to maintain

Best Buy Bring creative new

solutions to market

faster, shorten the

learning curve, lower

costs

Portal (RetailZone),

Employee Toolkit,

Communities of Practice

(Retail and Services)

1.5 percent increase in gross

margin

Sold 4.2 units/store/day more in

pilot stores

3 percent drop in damage claims

Paper reduction savings of

$250K/year

Total investment of

approximately $3.5 million

Data gleaned from APQC’s “Successfully Implementing KM” (1999), “Managing Content and Knowledge” (2001), and “Retaining

Valuable Knowledge” (2002) benchmarking studies

OrganizationOrganizationOrganization Target Value

Proposition

Target Value

Proposition

Target Value

Proposition ApproachApproachApproach ResultsResultsResults

ChevronTexacoChevronTexaco Reduce operating costs,

improve operational

excellence, improve

safety

Reduce operating costs,

improve operational

excellence, improve

safety

CoPs, facilitate transfer of

best practices, People Finder

CoPs, facilitate transfer of

best practices, People Finder Two billion dollar reduction in

annual operating costs (1991 v.

1998); $670 million came from

refining best practices.

Total investment of more than

$2 million (total figure unknown)

Two billion dollar reduction in

annual operating costs (1991 v.

1998); $670 million came from

refining best practices.

Total investment of more than

$2 million (total figure unknown)

Dow ChemicalDow Chemical Provide faster access to

information, improve

information

management, improve

sales leads

Provide faster access to

information, improve

information

management, improve

sales leads

Content management,

communities of practice

Content management,

communities of practice Increased number of sales leads

—Increase in new product sales

Improved customer satisfaction

scores

CM investment of over $3

million for start up, $8 million

annually.

Increased number of sales leads

—Increase in new product sales

Improved customer satisfaction

scores

CM investment of over $3

million for start up, $8 million

annually.

GE PlasticsGE Plastics Decrease customer

service costs

Decrease customer

service costs Customer portal, customer

knowledge repository

Customer portal, customer

knowledge repository Number of test chips created

decreased from 4.2 to 2.7

Average reduction of 4.5 hours per

color match

Savings of $2.25 million per year

Total investment unknown

Number of test chips created

decreased from 4.2 to 2.7

Average reduction of 4.5 hours per

color match

Savings of $2.25 million per year

Total investment unknown

ShellShell Create a single, global

company

Reduce cycle time

“Too fast to follow”

Create a single, global

company

Reduce cycle time

“Too fast to follow”

Global Networks (CoPs)

New ways of working

Letting the new guys into “Old

Boy” networks

Transfer of best practices

Global Networks (CoPs)

New ways of working

Letting the new guys into “Old

Boy” networks

Transfer of best practices

$200 million/yr cost savings

Reduced number of wells

Increased facility uptime

Reduced design and planning

errors

Total investment of

approximately $4 million

$200 million/yr cost savings

Reduced number of wells

Increased facility uptime

Reduced design and planning

errors

Total investment of

approximately $4 million

BPBP Know-how:

A brand attribute;

ability to innovate and

execute faster and

smarter than competitors

Know-how:

A brand attribute;

ability to innovate and

execute faster and

smarter than competitors

Networks, Peer Assist, AARs,

Retrospects, Technology VP

support, Operations Value

Process

Networks, Peer Assist, AARs,

Retrospects, Technology VP

support, Operations Value

Process

$260 million cost savings

Drilling cost reduction for

Schiehallion

(West of Shetlands)

Refinery turnarounds

Retail site construction , Total

investment unknown

$260 million cost savings

Drilling cost reduction for

Schiehallion

(West of Shetlands)

Refinery turnarounds

Retail site construction , Total

investment unknown

SchlumbergerSchlumberger Knowledge in the hands

of employees and

customers

Knowledge in the hands

of employees and

customers

CoPs, InTouch KM system,

intranet, extranet, content

management

CoPs, InTouch KM system,

intranet, extranet, content

management

$200 million cost savings

95 percent reduction in time to

resolve tech queries

75 percent reduction in updating

modifications

Total investment of

approximately $20 million

$200 million cost savings

95 percent reduction in time to

resolve tech queries

75 percent reduction in updating

modifications

Total investment of

approximately $20 million

Cap Gemini Ernst &

Young

Cap Gemini Ernst &

Young Faster revenue growth,

lower costs

Faster revenue growth,

lower costs CoPs, central KM managers,

content management

CoPs, central KM managers,

content management Ten-fold increase in revenue with

only five-fold increase in

employees

Ten-fold increase in revenue with

only five-fold increase in

employees

IBM Global ServicesIBM Global Services Revenue growth,

industry leadership

Revenue growth,

industry leadership COPs, knowledge managers,

Intellectual Capital

Management System

COPs, knowledge managers,

Intellectual Capital

Management System

400 percent increase in service

revenue

Time savings of $24 million in

1997

Approximately $750K to start

up, $750K annually to maintain

400 percent increase in service

revenue

Time savings of $24 million in

1997

Approximately $750K to start

up, $750K annually to maintain

Best BuyBest Buy Bring creative new

solutions to market

faster, shorten the

learning curve, lower

costs

Bring creative new

solutions to market

faster, shorten the

learning curve, lower

costs

Portal (RetailZone),

Employee Toolkit,

Communities of Practice

(Retail and Services)

Portal (RetailZone),

Employee Toolkit,

Communities of Practice

(Retail and Services)

1.5 percent increase in gross

margin

Sold 4.2 units/store/day more in

pilot stores

3 percent drop in damage claims

Paper reduction savings of

$250K/year

Total investment of

approximately $3.5 million

1.5 percent increase in gross

margin

Sold 4.2 units/store/day more in

pilot stores

3 percent drop in damage claims

Paper reduction savings of

$250K/year

Total investment of

approximately $3.5 million

Data gleaned from APQC’s “Successfully Implementing KM” (1999), “Managing Content and Knowledge” (2001), and “Retaining

Valuable Knowledge” (2002) benchmarking studies

Data gleaned from APQC’s “Successfully Implementing KM” (1999), “Managing Content and Knowledge” (2001), and “Retaining

Valuable Knowledge” (2002) benchmarking studies