Process Change Patterns: Recent Research, Use Cases, Research Directions [original]

Process Change Patterns:

Recent Research, Use Cases, Research Directions

Manfred Reichert1and Barbara Weber2

1University of Ulm, Germany, [email protected]

2University of Innsbruck, Austria, [email protected]

Abstract. In previous work, we introduced change patterns to foster

a systematic comparison of process-aware information systems with re-

spect to change support. This paper revisits change patterns and shows

how our research activities have evolved. Further, it presents character-

istic use cases and gives insights into current research directions.

1 Introduction

Information systems (IS) are increasingly aligned in a process-oriented way. This

emerging generation of IS is referred to as process-aware information systems

(PAIS) [1]. A PAIS should support real-world processes properly, i.e., there

should be no mismatch between the processes implemented by it and those ex-

isting in reality. Hence, advanced support is needed for customizing a PAIS to its

application environment as well as for quickly adapting implemented processes

to changing needs. The increasing demand for process change support poses new

challenges for IS engineers and requires the use of change enabling technologies.

Accordingly, a method is required that allows PAIS engineers to systemati-

cally assess the change capabilities of available technologies. In [2], we introduced

change patterns as well as change support features to enable such a systematic

assessment of PAIS with respect to process change support. In particular, change

patterns allow for high-level process adaptations. In turn, change support fea-

tures summarize fundamental features to be provided by a PAIS in order to

change and evolve implemented processes in a correct, robust and secure way.

This paper discusses how our research on change patterns has evolved, how

they have been used in theory and practice, and what research directions are.

2 Background: Process Change Patterns

Originally, in [2] we introduced 17 patterns for realizing control flow changes.

These patterns reduce the complexity of process changes and raise the level for

expressing changes by providing abstractions above the level of primitive change

operations. To structure the change patterns, we divided them into adaptation

patterns and change patterns for predefined changes (cf. Fig. 2 in [2]). While the

former enable structural changes of a process schema, the latter allow process

participants to add information regarding unspecified parts of a process schema

during run-time.

An adaptation pattern (AP) enables structural changes of process schemes.

AP1 (AP2) allows inserting (deleting) a process fragment. Moving and replac-

ing fragments is supported by AP3 (Move Process Fragment), AP4 (Replace

Process Fragment), AP5 (Swap Process Fragment), and AP14 (Copy Process

Fragment). AP6 and AP7 allow adding or removing levels of hierarchy: the ex-

traction of a sub-process from a process schema is supported by AP6, whereas the

inclusion of a sub-process into a process schema is supported by AP7. Patterns

AP8–AP12 support adaptations of control dependencies: embedding a process

fragment in a loop (AP8), parallelizing a process fragment (AP9), embedding

a process fragment in a conditional branch (AP10), and adding / removing

control dependencies (AP11, AP12). Finally, AP13 allows changing transition

conditions. Generally, the region to which an adaptation pattern is applied may

be chosen dynamically. Hence, adaptation patterns are well suited for realizing

ad-hoc changes and coping with the evolving nature of business processes [1]. For

each adaptation pattern, we have provided a name, a description, an illustrating

example, a description of the problem it addresses, a couple of design choices,

remarks regarding its implementation, and references to related patterns. In this

context, design choices allow parameterizing change patterns keeping the number

of distinct patterns manageable.

Patterns for changes in pre-defined regions allow for better dealing with

uncertainty by deferring decisions regarding the exact control-flow to the run-

time. Instead of requiring a process model to be fully specified prior to execution,

parts of the model may remain unspecified. As opposed to adaptation patterns,

whose application is not restricted a priori to a particular process part, the

parts of a process schema that may be changed or expanded are constrained.

In this category, we identified 4 patterns: Late Selection (PP1), Late Modeling

(PP2) and Late Composition of Process Fragments (PP3), and Multi-Instance

Activity (PP4). These four patterns differ regarding the parts that may remain

unspecified resulting in a different degree of freedom during run-time.

3 Recent Research on Process Change Patterns

In recent work we have detailed the change patterns and provided empirical

evidence for them. Further, we have formalized and implemented them. In detail:

Detailing change patterns and empirical evidence. We extended our orig-

inal work in [3], which provides an in-depth description of all change patterns;

it describes the pattern selection criteria, the data sources used, and the pro-

cedure applied for pattern identification. Further, it discusses how the patterns

were identified based on the analysis of large process model collections from the

healthcare and automotive domains. Finally, [3] introduces additional patterns

and provides an extended pattern-based evaluation of selected approaches from

industry as well as academia.

Change pattern formalization. To obtain unambiguous pattern descriptions

and ground pattern implementation as well as pattern-based analyses on a sound

basis, we provided a formal semantics for change patterns in [4]. For each change

pattern, its formal semantics is specified by comparing the execution traces pro-

ducible on a process schema before and after applying the pattern to it. The

formalization is independent from any process meta model and thus allows im-

plementing the patterns in a variety of process support tools.

Pattern implementation. The change patterns were implemented in an adap-

tive PAIS – the AristaFlow BPM Suite [5]. The adaptation patterns are realized

in terms of high-level change operations, which can be used for creating and

changing process schemes. Hence, flexible exception handling and controlled pro-

cess evolution become possible. Further, adaptation patterns are associated with

pre-/post-conditions to ensure structural and behavioral soundness of a process

schema after pattern application; i.e., correctness by construction is ensured.

Recently, we complemented the existing workflow patterns by a set of time

patterns to make PAIS comparable with respect to their ability to deal with

temporal constraints [6].

4 Characteristic Use Cases for Change Patterns

On one hand, change patterns provide the basis for realizing changes in different

stages of the process life cycle [7]. On the other, they serve as benchmark for

evaluating change support in existing languages and tools.

4.1 Supporting Process Changes Along the Process Life Cycle

We first discuss fundamental use cases for realizing changes in different stages

of the process life cycle:

Process schema creation. Change patterns have been used for intelligent

process schema creation [8]. For example, AristaFlow allows modeling a sound

process schema based on an extensible set of adaptation patterns [5]. Only those

patterns may be applied in a given context, which do not violate the soundness

of the process schema. In turn, [9] describes a set of pattern compounds, simi-

lar to the adaptation patterns, allowing for the context-sensitive selection and

composition of workflow patterns during process modeling. Finally, adaptation

patterns have been used for the model-based integration of services into business

applications at later stages during the process life cycle [10].

Process schema configuration. The configuration of a reference process schema

constitutes another use case for adaptation patterns. Provop, for example, allows

creating a process variant by applying a sequence of adaptation patterns (e.g.,

AP1, AP2 or AP3) to the given reference schema [11]. By utilizing the semantics

of the adaptation patterns applied in a given configuration setting, it is further

ensured that the resulting process variant schema is sound [12].

Process instance change. An important use case is to enable actors to deviate

from a pre-specified process schema at run-time, e.g., to cope with exceptions.

For this purpose, AristaFlow supports instance-specific changes based on the

same adaptation patterns as used for process modeling [5]. Further, it utilizes the

semantics of the applied adaptation patterns to ensure correctness of the result-

ing process instance schema. Recently, approaches aiming at automated instance

changes have emerged. Usually, they only consider a subset of the adaptation

patterns. For example, Q-Advice uses AP1 and its variants to automatically

inject quality measure activities into the workflows of software engineers at run-

time. The activities to be added are determined situationally using contextual

knowledge and quality goal tracking [13]. A more generic approach to automate

instance adaptations, which is based on declarative processes and planning, is

described in [14]. Regarding ad-hoc changes, [15] additionally ensures compliance

of process instance adaptations with defined semantic constraints. For this, inte-

ger programming formulation is used to validate the applied adaptation patterns

against the given set of semantic constraints (AP1–AP5 are considered). An ap-

proach for the flexible support of product development processes is presented in

[16]; the sub-processes of such a process, which refine analysis, synthesis, and

verification activities, may be dynamically selected to allow for a flexible pro-

cess execution without need for structural adaptations. Thereby, a subset of the

patterns for changes in pre-defined regions is considered (i.e., PP1–PP3).

Process schema evolution. Adaptive PAIS allow for schema evolution consid-

ering version management and the migration of already running process instances

to the new schema version. [17] presents techniques for detecting and resolving

conflicting change operations, which rely on selected adaptation patterns and

their semantics. In turn, [18] shows how to compute a sequence of adaptation

patterns required to transform a given schema version into another one. Both

scenarios consider AP1, AP2, and AP3. Particularly, adaptation patterns play a

crucial role for ensuring the correctness of schema changes and instance migra-

tions. In AristaFlow, schema evolution is based on the same adaptation patterns

as used for process modeling and ad-hoc changes [5]. Thereby, pattern seman-

tics is utilized to cope with conflicting changes at the type and instance level,

to increase the number of migratable process instances, and to efficiently repre-

sent applied changes [19–21]. Note that similar concepts exist for evolving ser-

vice compositions [22]. Furthermore, continuous process improvement, relying on

case-based reasoning and adaptation patterns, is considered in [23]. Finally, [24]

introduces patterns for co-evolving processes and software architectures. These

patterns are based on selected adaptation patterns and allow describing the im-

pact a business process change has on corresponding software architectures.

Process schema refactoring. A specific kind of schema evolution is provided

by process schema refactorings; i.e., syntactical transformations of a process

schema not changing its behavior. Examples of such refactorings and their rela-

tion to adaptation patterns (e.g., AP6 and AP7) are discussed in [25].

Process change reuse. When handling exceptions, it might be useful to reuse

changes applied in similar problem contexts earlier [26]. For example, ProCycle

fosters change reuse based on case-based reasoning and semantic change anno-

tations [7]. Further, it supports AP1–AP5 and utilizes their specific semantics

to adjust parameter settings of recorded changes when reusing them.

Process schema comparison. Comparing two process schemes is crucial to

decide how similar the schemes are or how to derive the one from the other. In

this context, adaptation patterns can be used to describe the structural difference

(i.e., edit distance) between schemes in terms of high-level changes. Based on

specific variants of patterns AP1–AP3, [27] presents a technique that allows

determining this difference. A similar approach is presented in [28].

Process change analysis. Adaptive PAIS capture process changes in change

logs, which record applied adaptation patterns and their parameter settings. For

change analysis, different techniques exist. Based on AP1-AP3, [29] applies pro-

cess mining to change logs to discover change processes providing an aggregated

visualization of all changes. In turn, MinAdept does not presume the existence

of a change log, but allows analyzing a collection of process variants derived from

the same schema [30]; algorithms are provided discovering a reference process

schema whose average edit distance to the process variants is minimal.

In summary, process change patterns are relevant for a variety of use cases in

the process life cycle. As shown, the patterns have served as basis for the design

and implementation of techniques supporting these use cases. While tools like

AristaFlow enable a broad support of most use cases and adaptation patterns,

other proposals only consider a specific use case and a subset of the adaptation

patterns.

4.2 Assessing and Designing Process Change Frameworks

Change patterns have been used for realizing pattern catalogs for specific mod-

eling languages, assessing existing PAIS, and enabling user-friendly changes. Ex-

amples of corresponding approaches are presented in the following.

Realizing a pattern catalog for a specific modeling language. [31] com-

bines change, exception and time patterns into a BPMN pattern catalog. Change

patterns are referred to as generic patterns, which are tailored and extended to

be applicable to BPMN. In turn, [32] uses the adaptation patterns for designing

a pattern catalog for BPEL schema changes.

Assessing existing approaches. A measure for a pattern-based assessment

of service orchestration languages is defined in [33]. In particular, the designed

pattern catalog includes the patterns for changes in predefined regions (i.e.,

PP1–PP4) and discusses how they are supported in existing BPEL dialects.

Enabling user-friendly changes. [34] presents an approach enabling end users

to change large process schemes based on personalized process views; AP1, AP2,

and AP8–AP10 may be applied to a process view, followed by the propagation

of the defined changes to the underlying process schema. In turn, [35] introduces

a user-centric approach for creating, changing and visualizing process schemes

based on the Concurrent Task Tree (CTT) – a task modeling language known

from end-user programming. Thereby, the described adaptation patterns are

mapped to CTT change operations. Finally, [36] presents a controlled experiment

that investigates the way users create and change process schemes on multi-touch

devices. Based on this, a gesture set for realizing adaptation patterns AP1, AP2,

AP6, AP7, AP8, AP10, and AP11 on multi-touch devices is designed.

5 Research Directions

When using change patterns for modeling, the quality of process schemes might

increase. Particularly appealing in this context is the mentioned correctness by

construction. However, the use of change patterns implies a different way of

creating process schemes compared to the use of change primitives. First of

all, the correctness-by-construction principle imposes a rather structured way of

modeling and hence constraints on change pattern combinations. In addition, the

exact set of change patterns (e.g., presence vs. non-presence of the move pattern)

might determine how patterns have to be combined to create a process fragment.

While the creation of process schemes based on change primitives has caused

attention in recent years [37], only little is known about the process of process

modeling when utilizing change patterns. To obtain an in-depth understanding

of it, we are currently working on empirical studies on the use of change patterns.

References

1. Reichert, M., Weber, B.: Enabling Flexibility in Process-Aware Information Sys-

tems - Challenges, Methods, Technologies. Springer (2012)

2. Weber, B., Rinderle, S., Reichert, M.: Change patterns and change support features

in process-aware information systems. In: Proc. CAiSE’07. (2007) 574–588

3. Weber, B., Reichert, M., Rinderle-Ma, S.: Change patterns and change support

features -enhancing flexibility in process-aware information systems. Data and

Knoweldge Engineering 66 (2008) 438–466

4. Rinderle-Ma, S., Reichert, M., Weber, B.: On the formal semantics of change

patterns in process-aware information systems. In: Proc. ER’08. (2008) 279–293

5. Dadam, P., Reichert, M.: The ADEPT project: a decade of research and develop-

ment for robust and flexible process support. Comp Scie - R&D 23 (2009) 81–97

6. Lanz, A., Weber, B., Reichert, M.: Time patterns for process-aware information

systems. Requirements Engineering (2013)

7. Weber, B., Reichert, M., Wild, W., Rinderle-Ma, S.: Providing integrated life

cycle support in process-aware information systems. Int’l Journal of Cooperative

Information Systems 18 (2009) 115–165

8. Smirnov, S., Weidlich, M., Mendling, J., Weske, M.: Object-sensitive action pat-

terns in process model repositories. In: Proc. BPM’10 Workshops. (2010) 251–263

9. Gschwind, T., Koehler, J., Wong, J.: Applying patterns during business process

modeling. In: Proc BPM’08. (2008) 4–19

10. Heller, M., Allgaier, M.: Model-based service integration for extensible enterprise

systems with adaptation patterns. In: ICE-B. (2010) 163–168

11. Hallerbach, A., Bauer, T., Reichert, M.: Capturing variability in business process

models: The Provop approach. Journal of Software Maintenance and Evolution:

Research and Practice 22 (2010) 519–546

12. Hallerbach, A., Bauer, T., Reichert, M.: Guaranteeing soundness of configurable

process variants in Provop. In: Proc. CEC’09. (2009) 98–105

13. Grambow, G., Oberhauser, R., Reichert, M.: Contextual injection of quality mea-

sures into software engineering processes. Int’l J Adv in Software 4(2011) 76–99

14. Marrella, A., Mecella, M., Russo, A.: Featuring automatic adaptivity through

workflow enactment and planning. In: Proc CollaborateCom’11. (2011) 372–381

15. Kumar, A., Yao, W., Chu, C.H., Li, Z.: Ensuring compliance with semantic

constraints in process adaptation with rule-based event processing. In: Proc

RuleML’10. (2010) 50–65

16. Reichel, T., R¨unger, G., Steger, D.: Flexible workflows for an energy-oriented

product development process. In: Proc ISSS/BPSC’10. (2010) 243–254

17. Gerth, C., Kster, J., Luckey, M., Engels, G.: Detection and resolution of conflicting

change operations in version management of process models. SOSYM (2011) 1–19

18. K¨uster, J., Gerth, C., Engels, G.: Dynamic computation of change operations in

version management of business process models. In: ECMFA’10. (2010) 201–216

19. Rinderle, S., Reichert, M., Dadam, P.: On dealing with structural conflicts between

process type and instance changes. In: Proc. BPM’04, Potsdam (2004) 274–289

20. Rinderle-Ma, S., Reichert, M., Weber, B.: Relaxed compliance notions in adaptive

process management systems. In: Proc. ER’08. (2008) 232–247

21. Rinderle, S., Reichert, M., Jurisch, M., Kreher, U.: On representing, purging, and

utilizing change logs in process management systems. In: BPM’06. (2006) 241–256

22. Andrikopoulos, V., Benbernou, S., Papazoglou, M.P.: Managing the evolution of

service specifications. In: Proc. CAiSE’08. (2008) 359–374

23. Kim, D., Lee, N., Kang, S.H.: An approach to continuous process improvement

based on case-based reasoning and process change patterns. IJICIC 8(2011)

24. Jamshidi, P., Pahl, C.: Business process and software architecture model co-

evolution patterns. In: Proc. MISE’12. (2012) 91–97

25. Weber, B., Reichert, M., Mendling, J., Reijers, H.A.: Refactoring large process

model repositories. Computers in Industry 62 (2011) 467–486

26. Aghakasiri, Z., Mirian-Hosseinabadi, S.H.: Workflow change patterns: Opportuni-

ties for extension and reuse. In: Proc. SERA’09. (2009) 265–275

27. Li, C., Reichert, M., Wombacher, A.: On measuring process model similarity based

on high-level change operations. In: Proc. ER’08. (2008) 248–264

28. K¨uster, J.M., Gerth, C., F¨orster, A., Engels, G.: Detecting and resolving process

model differences in the absence of a change log. In: BPM’08. (2008) 244–260

29. G¨unther, C.W., Rinderle, S., Reichert, M., van der Aalst, W.M.P.: Change mining

in adaptive process management systems. In: Proc. CoopIS’06. (2006) 309–326

30. Li, C., Reichert, M., Wombacher, A.: Mining business process variants: Challenges,

scenarios, algorithms. Data & Knowledge Engineering 70 (2011) 409–434

31. D¨ohring, M., Zimmermann, B., Karg, L.: Flexible workflows at design- and runtime

using bpmn2 adaptation patterns. In: Proc. BIS. (2011) 25–36

32. Tragatschnig, S., Zdun, U.: Runtime process adaptation for bpel process execution

engines. In: EDOCW, IEEE Computer Society (2011) 155–163

33. Lenhard, J., Sch¨onberger, A., Wirtz, G.: Edit distance-based pattern support

assessment of orchestration languages. In: OTM Conferences (1). (2011) 137–154

34. Kolb, J., Kammerer, K., Reichert, M.: Updatable process views for user-centered

adaption of large process models. In: Proc. ICSOC’12. (2012) 484–498

35. Kolb, J., Reichert, M., Weber, B.: Using concurrent task trees for stakeholder-

centered modeling and visualization of business processes. In: S-BPM ONE. (2012)

36. Kolb, J., Rudner, B., Reichert, M.: Towards gesture-based process modeling on

multi-touch devices. In: Proc. CAiSE Workshops. (2012) 280–293

37. Pinggera, J., et al: Modeling styles in business process modeling. In: BM-

MDS/EMMSAD. (2012) 151–166