Providing Support for the Optimized Management of Declarative Processes [original]

Providing Support for the Optimized Management of

Declarative Processes

Irene Barba1, Andreas Lanz2Andr´

es Jim´

enez-Ram´

ırez1, Barbara Weber3, Manfred

Reichert2, and Carmelo Del Valle1

1Departamento de Lenguajes y Sistemas Inform´

aticos, University of Seville, Spain,

{irenebr,ajramirez,carmelo}@us.es

2Institute of Databases and Information Systems, Ulm University, Germany

{Andreas.Lanz,Manfred.Reichert}@uni-ulm.de

3Department of Computer Science, University of Innsbruck, Austria,

Technical University of Denmark, Denmark

[email protected]

Abstract. Declarative process models are becoming increasingly popular due

to the high flexibility they offer to process participants. Based on a declarative

process model, there exist numerous possible enactment plans, each one with

specific values for relevant objective functions (e.g., overall completion time).

How to actually execute such a model is quite challenging due to several reasons:

(1) proper objective functions must be considered to find optimized enactment

plans, (2) users often do not have an understanding of the overall process, (3) the

presence of a variety of temporal constraints to be met during process enactment,

and (4) the need to coordinate multiple instances of a process concurrently exe-

cuted (which compete for shared resources). This is further complicated by the

fact that the enactment of new process instances may continuously start over time

and many organizations do not exactly know their future demands. In such con-

text, to properly support users in enacting declarative process models, this paper

suggests generating optimized enactment plans from declarative process models.

The generated enactment plans may be used for different purposes, e.g., to pro-

vide personal schedules to users. Moreover, they may be dynamically adapted if

required. To evaluate the applicability of our approach in practical settings we

apply it to a real process scenario from the healthcare domain.

Keywords: Process flexibility, declarative process model, temporal constraints,

constraint programming, scheduling, healthcare processes

1 Introduction

For several years, there has been an increasing interest in aligning information systems

in a process-oriented way [2, 15]. Usually, business processes (BPs) have numerous

constraints to be obeyed during process enactment. To provide operational support,

BPs are mostly specified in an imperative way, i.e., defining a precise schema that es-

tablishes how a given set of activities has to be performed. However, imperative process

models are often too rigid to meet the flexibility requirements of users. As an alternative,

therefore, declarative process models are increasingly used [13, 9] since they provide an

obtains

Constraint-

based Approach

Personal Schedules

Client Appointments

Time Prediction

Optimized

Enactment

Plans

TConDec-R

Specification

Process

Information

...

replanning

abc

Selected

Plan

selects

updates

supports

models

Fig. 1. Overview of our approach.

increased flexibility when executing the process, allowing users to specify what has to

be done instead of how [8].

On one hand, a declarative process model offers a high flexibility to end users by al-

lowing for the required degree of freedom. On the other, executing a declarative model

usually entails larger efforts for users compared to imperative models [10]. In general,

numerous enactment plans related to the same declarative model exist, each one pre-

senting specific values for relevant objective functions (e.g., overall completion time).

Consequently, the decision on how to execute a declarative model is quite complex. It

becomes even more challenging when dealing with business processes that (1) contain

complex temporal as well as cross-instance constraints, (2) require an efficient manage-

ment of shared resources, and (3) need to consider the optimization of specific objective

functions in an uncertain and changing environment. In such a context, usually, users

neither know the complete BP nor the impact of their actions. Both might result in sub-

optimal enactment plans. Therefore, advanced user support is required when executing

declarative process models. Such support can be provided by automatically generating

optimized enactment plans from the respective declarative process model. Moreover,

during enactment it should be possible to flexibly adapt these plans if required taking

relevant run-time information into account.

Optimized enactment plans foster sophisticated user support through: (1) personal

schedules [4], i.e., assisting process participants by recommending which activities they

shall perform when, (2) reduced turnaround and waiting times, i.e., organizing appoint-

ments while taking the start times of the activities the users are involved in into account,

and (3) time predictions, i.e., predicting enactment times for future activities. Note that

(1) allows for a better planning of work, while (3) enables decision support for process

participants [14].

2 Contribution

This paper suggests the generation of optimized enactment plans for a set of process

instances derived from a declarative model. Figure 1 provides an overview of our ap-

proach. First, the declarative specification of the BP is defined (Step a in Fig. 1). For

this, we consider the declarative language DECLARE1[13] as basis. In order to ad-

dress the described problems, DECLARE is extended by (1) including capabilities for

reasoning about resources as well as the estimated duration of process activities, and

1DECLARE is one of the most referenced and used declarative process modeling languages.

(2) supporting common time patterns reported in literature [6, 5] (e.g., cross-instance

constraints related to time). This results in the TConDec-R language.

From a TConDec-R specification, in turn, optimized enactment plans can be auto-

matically generated (Step b in Fig. 1). In this context, the activities to be executed are

selected and ordered (i.e., planning problem [3]), considering control-flow constraints

as well as temporal and resource constraints imposed by the declarative specification

(i.e., scheduling problem [1]). For planning and scheduling the activities in a way such

that a given objective function becomes optimized (i.e., for generating a set of opti-

mized enactment plans), a constraint-based approach can be proposed. Finally, from

the set of optimized enactment plans (i.e., schedules), the plan that fits best to the sce-

nario requirements is selected for enactment (Step c in Fig. 1). The selected plan is then

used for improving process support in several respects (Step f in Fig. 1).

Additionally, the proposed approach supports the dynamic adaptation of optimized

enactment plans during run-time through replanning, and hence enables an increased

run-time flexibility (Step d in Fig. 1). Each time the set of optimized enactment plans

is updated, a new plan for providing process support is selected. In general, replanning

might become necessary either if the actual enactment of the process instances deviates

from the optimized enactment plan generated earlier (e.g., estimates might turn out

to be inaccurate or resource availability might unexpectedly change) or in scenarios in

which the enactment of new process instances continuously starts and the demand of the

services to be provided varies over time. Respective scenarios are known as lot-sizing

ones with uncertain demand [12].

To evaluate the applicability of our approach in practical settings we consider to

apply it to a real process scenario from the healthcare domain. The considered scenario

deals with the scheduling of surgeries and their preparations in the context of ovarian

carcinoma [11, 7]. The application of the proposed approach to the medical scenario

offers several advantages: (1) avoiding constraint violations, (2) managing shared re-

sources in an effective way, (3) improving and automating (inter-process) coordination

between different hospitals, (4) reducing waiting times of patients as the schedule be-

comes known beforehand, and (5) minimizing the length of patient stays in hospitals.

Note that the approach is not restricted to healthcare environments, i.e., it can be

also applied in other domains for which process flexibility and temporal constraints

play a crucial role (e.g., automotive engineering and flight planning [6]).

The proposed approach combines original contributions regarding (1) the formal

specification and operational support of complex temporal constraints during process

enactment, (2) cross-instance coordination, (3) resource allocation (i.e., scheduling) be-

fore and during process enactment, and (4) optimization of objective functions.

3 Conclusion

This paper proposed the generation of optimized enactment plans (e.g., minimizing

overall completion time) from declarative temporal process models. This generation is

quite challenging since it considers complex temporal constraints, cross-instance coor-

dination, resource allocation before and during process enactment, and optimization of

a given objective function. It is further complicated by the fact that the enactment of

new process instances may continuously start over time and many organizations do not

exactly know their future demands.

The generated plans can be used for different purposes, e.g., providing users with

a personal schedule, suggesting appointments to process stakeholders, or predicting

enactment times for process activities. Moreover, these plans are adapted during the

enactment of the process if necessary.

As future work, we will test our approach in the context of ovarian carcinoma.

References

1. P. Brucker and S. Knust. Complex Scheduling (GOR-Publications). Springer, 2006.

2. M. Dumas, W.M.P. van der Aalst, and A.H.M. ter Hofstede. Process-Aware Information

Systems: Bridging People and Software through Process Technology. Wiley-Interscience,

Hoboken, NJ, 2005.

3. M. Ghallab, D. Nau, and P. Traverso. Automated Planning: Theory and Practice. Morgan

Kaufmann, Amsterdam, 2004.

4. A. Lanz, J. Kolb, and M. Reichert. Enabling personalized process schedules with time-aware

process views. In CAiSE’13 Workshops, LNBIP, pages 205–216. Springer, 2013.

5. A. Lanz, M. Reichert, and B. Weber. Process time patterns: A formal foundation. Information

Systems, 57:38–68, 2016.

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

Requirements Engineering, 19(2):113–141, 2014.

7. Ovarian cancer (CG122). http://www.nice.org.uk/CG122, 2011. [Online; accessed 7-

April-2016].

8. M. Pesic. Constraint-Based Workflow Management Systems: Shifting Control to Users. PhD

thesis, Technische Universiteit Eindhoven, The Netherlands, 2008.

9. P. Pichler, B. Weber, S. Zugal, J. Pinggera, J. Mendling, and H.A. Reijers. Imperative ver-

sus Declarative Process Modeling Languages: An Empirical Investigation. In Proc. BPM

Workshops, pages 383–394, 2011.

10. M. Reichert and B. Weber. Enabling Flexibility in Process-Aware Information Systems.

Springer, 2012.

11. B. Schultheiß, J. Meyer, R. Mangold, T. Zemmler, and M. Reichert. Designing the processes

for ovarian cancer surgery (in german). Technical Report DBIS-6, University of Ulm, 1996.

12. L. Tiacci and S. Saetta. Demand forecasting, lot sizing and scheduling on a rolling horizon

basis. International Journal of Production Economics, 140(2):803–814, 2012.

13. W.M.P. van der Aalst, M. Pesic, and M.H. Schonenberg. Declarative workflows: Balancing

between flexibility and support. Computer Science - Research and Development, 23(2):99–

113, 2009.

14. W.M.P. van der Aalst, M.H. Schonenberg, and M. Song. Time prediction based on process

mining. Information Systems, 36(2):450–475, 2011.

15. M. Weske. Business Process Management: Concepts, Languages, Architectures (Second

Edition). Springer, 2012.