Work ow Modelling Assistance by Means of Process-oriented Case-based Reasoning

Abstract

Copyright c 2016 for this paper by its authors. Copying permitted for private and academic purposes. In Proceedings of the ICCBR 2016 Workshops. Atlanta, Georgia, United States of America

-

In recent years, work ows have become an important paradigm to represent processes in many application areas. Traditionally known from the eld of business processes, work ows are \the automation of a business process, in whole or part, during which documents, information or tasks are passed from one participant to another for action, according to a set of procedural rules" [ 25 ]. Today, their application has extended towards various other domains such as healthcare [ 7 ], or the analysis of large data sets employing so called scienti c work ows [ 24 ]. Moreover, they are considered as a novel programming paradigm [ 14 ] and can also be used to represent simple processes like cooking instructions [ 22 ].

In all these domains, the creation of work ows (also refered to as work ow modelling) is a complex and time consuming task. Consequently, in order to ease their creation, reuse becomes an important key to the successful application of the work ow paradigm. Work ow reuse is usually supported by search methods capturing the current needs and requirements in a query or a partial work ow. Based on this, a previously modelled work ow is identi ed which at best matches the current scenario. Thus, the work ow does not have to be modelled from scratch. Various approaches already exist for such search-based work ow reuse (e.g., [ 5,8,2,3 ]).

However, work ow adaptations are required more often due to an increased individuality demand of work ows, i.e., work ows need to be tailored to the particular needs or scenario more frequently. Manual work ow adaptation may become an elaborate task. Thus, methods supporting work ow adaptations are of high relevance. In this regard, Process-oriented Case-based Reasoning (POCBR) [ 13 ] can be a means to support the creation and adaptation of processes that are, e.g., represented as work ows. Although, POCBR is of high relevance little research exists so far, in particular addressing work ow adaptation (e.g., [ 12 ]).

The presented approach addresses work ow modelling assistance by means of POCBR, in particular focussing on the adaptation of work ows. Adaptation requires knowledge, e.g., in the form of adaptation rules. The modelling of such domain speci c adaptation knowledge, however, is expensive and requires a deep understanding of the adaptation methods. This results in an acquisition bottleneck for adaptation knowledge [ 9 ]. Thus, the approach includes automated learning of adaptation knowledge to prevent limited adaptation capabilities. 1

Approach

The overall approach for work ow modelling support by means of ProcessOriented Case-based Reasoning is illustrated in Figure 1. In order to determine the requirements and needs on the work ow to be modelled, the query language POQL [ 19 ] basically captures work ow elements or subwork ows that are desired or undesired. Thus, the best matching work ow from a case base (or work ow repository respectively) of previously modelled work ows can be identi ed. Next, work ow adaptations are executed in order to increase the ful lment of the adapted work ow w.r.t. the POQL query. Hence, the retrieved work ow W is transformed into an adapted work ow Wn by chaining various adaptation steps W !s1 W1 !s2 : : : !sn Wn. This process can in principle be considered a search process towards an optimal solution w.r.t. the query.

? POQL Query retrieval adaptation

specialization work owfrom generalizedcasebase adaptedgeneralizedwork ow adaptedwork ow casebase generalize casebase

learn adaptation generalized knowledge case base work ow streams adaptation operators

These adaptation steps are performed by di erent adaptation approaches, which are integrated in a single process such that they can be executed in alliance in order to improve adaptation capabilities [ 16 ]. More precisely, compositional adaptation [ 15 ] primarily replaces meaningful subcomponents of a work ow, called work ow streams, by more appropriate subcomponents of other work ows. Subsequently, adaptation operators [ 18 ], which are learned automatically from the case base, enable to remove, insert, or replace smaller work ow fragments. This adaptation process is further enhanced by generalization of work ows and adaptation knowledge [ 17 ]. Both consist of generalized work ow elements that are placeholders representing a set of possible elements. The adapted work ow containing such elements, is nally specialized according to the given query, i.e., suitable work ow elements are chosen.

Consequently, adaptation is executed based on the best maching work ow from the repository. However, Smyth and Keane [ 23 ] already stated, that it is important to re ect the adaptability during retrieval. Otherwise, retrieval may provide a work ow that cannot be at best adapted according to the query. Thus, the optimal work ow solution w.r.t. the query cannot always be ensured. Hence, there is a demand for methods enabling the assessment of adaptability of work ows, for example, by executing various example adaptations [ 6 ]. The main assumption is that work ows that are very adaptable in many scenarios are also highly adaptable in other scenarios. The computed adaptability value of the work ow can then be considered during retrieval. Another threat for retrieval and adaptation of work ows are insu ciently de ned work ow models in the repository. Such a lack of information can result in inappropriate work ows selected during retrieval or incompletely generated adaptation knowledge. Both hamper the presented work ow modelling assistance. To prevent this, work ow completion [ 20 ] aims at deriving the missing information automatically prior to retrieval and adaptation.

The modelling assistance has been implemented and evaluated using the CAKE (Collaborative Agent-based Knowledge Engine) framework 1 developed at the University of Trier [ 4 ]. The CookingCAKE2 prototype [ 16 ] demonstrates the presented approach in the cooking domain, where work ows represent cooking instructions consisting of preparation steps and involved ingredients. 2

Future Work

Future work comprises approaches for supporting the manual adaptation of work ows by means of the developed POCBR adaptation methods. A drawback of applying adaptation methods automatically, is that the adaptation goal must mostly be known previously. Consequently, this can lead to a non-optimal or undesired solution. Hence, user interaction [ 1 ] is a promising approach to overcome this drawback. Interactive adaptation could support the search of a suitable query and hence the desired solutions. The idea for supporting work ow adaptation is that the user is involved directly during adaptation by manipulating the work ow manually. Based on the users' work ow manipulations, the query can be re ned automatically. Subsequently, suggestions are made about a possible automatic work ow adaptation. This suggested adaptation can be accepted or rejected by the user. This process, helps to re ne the query stepwise and overcomes the drawback of an extensive search for the desired solution.

Finally, the developed methods will be analysed by an extensive summative evaluation. While the adaptation approaches have already been assessed seperately, an overall evaluation is planned to enable a comparison between the presented approaches. Main factors a ecting the utility of work ow adaptations are the increased ful lment of the work ow after adaptation w.r.t. the query as well as the quality of the adapted work ow. Thereto, quality metrics will be based on various quality frameworks [ 11,21,10 ]. 1 cakeflow.wi2.uni-trier.de 2 http://cookingcake.wi2.uni-trier.de

An analytical evaluation will provide information about several computable quality metrics such as the complexity of the work ow, the throughput time of the work ow as well as the ful llment of the query. Further, an experimental evaluation will be used to capture a more user-oriented view on the work ow quality and further aims at investigating the utility of adapted work ows. The basic approach is, that the work ow quality will be assessed by a blind experiment ignoring the query in which the users rate the adapted work ow and the retrieved work ow, not knowing which of these work ows have been adapted. Then, they rate several items on a likert scale to acquire users perceptions on the work ow models quality. Thus, the quality of the retrieved and adapted workows can be compared and conclusions on the impact of the work ow adaptation methods on the work ow quality can be drawn. Moreover, the utility of work ow adaptation methods is investigated. Since utility depends on the current goals [ 21 ], work ow utility can only be assessed knowing the query. Then, users rate whether they prefer the adapted or the retrieved work ow.

All work ow adaptation methods will be investigated in a joint as well as in separate applications. Thus, a comparison of the di erent adaptation methods can be achieved and conclusions on the overall adaptation process can be made.

References

1. Aha , D.W. , Mun~oz- Avila , H.: Introduction: Interactive case-based reasoning . Appl. Intell . 14 ( 1 ) ( 2001 )

2. Awad , A. : BPMN-Q: A language to query business processes . In: Reichert, M. , Strecker , S. , Turowski , K . (eds.) Enterprise Modelling and Information Systems Architectures - Concepts and Applications , Proceedings of EMISA'07 , St . Goar, Germany, 2007 . LNI, vol. P-119. GI ( 2007 )

3. Becker , M. , Laue , R.: A comparative survey of business process similarity measures . Computers in Industry 63 ( 2 ) ( 2012 )

4. Bergmann , R. , Gessinger , S. , Gorg, S. , Muller, G.: The collaborative agile knowledge engine CAKE . In: Goggins, S.P. , Jahnke , I. , McDonald , D.W. , Bj rn, P. (eds.) Proceedings of GROUP'14, Sanibel Island , FL, USA, 2014 . ACM ( 2014 )

5. Bergmann , R. , Gil , Y. : Similarity assessment and e cient retrieval of semantic work ows . Inf. Syst . 40 ( 2014 )

6. Bergmann , R. , Muller, G., Zeyen , C. , Manderscheid , J.: Retrieving adaptable cases in process-oriented case-based reasoning . In: Markov, Z. , Russell , I. (eds.) Proceedings of FLAIRS 2016 , Key Largo , Florida, 2016 . AAAI Press ( 2016 )

7. Dadam , P. , Reichert , M. , Kuhn , K. : Clinical Work ows - The Killer Application for Process- oriented Information Systems? Springer, London ( 2000 )

8. Dijkman , R.M. , Dumas , M., van Dongen , B.F. , Ka  arik , R., Mendling , J.: Similarity of business process models: Metrics and evaluation . Inf. Syst . 36 ( 2 ) ( 2011 )

9. Hanney , K. , Keane , M.T.: The adaptation knowledge bottleneck: How to ease it by learning from cases . Springer Berlin Heidelberg ( 1997 )

10. Jansen-Vullers , M. , Loosschilder , M. , Kleingeld , P. , Reijers , H.: Performance measures to evaluate the impact of best practices . In: Proceedings of Workshops and Doctoral Consortium of the 19th International Conference on Advanced Information Systems Engineering (BPMDS workshop) . vol. 1 . Tapir Academic Press Trondheim ( 2007 )

11. Lindland , O.I. , Sindre , G., S lvberg, A.: Understanding quality in conceptual modeling . IEEE Software 11(2) ( 1994 )

12. Minor , M. , Bergmann , R. , Gorg, S. , Walter , K. : Towards case-based adaptation of work ows . In: Bichindaritz, I. , Montani , S. (eds.) Case-Based Reasoning. Research and Development, ICCBR 2010 , Alessandria, Italy, July 19-22 , 2010 . Proceedings. LNCS, vol. 6176 . Springer ( 2010 )

13. Minor , M. , Montani , S. , Recio-Garc a , J.A.: Process-oriented case-based reasoning . Inf. Syst . 40 ( 2014 )

14. Morrison , J.P. : Flow-Based programming: A new approach to application development . CreateSpace , Paramount, CA ( 2010 )

15. Muller, G., Bergmann , R.: Work ow streams: A means for compositional adaptation in process-oriented CBR . In: Lamontagne, L. , Plaza , E. (eds.) Case-Based Reasoning Research and Development, ICCBR 2014 , Cork, Ireland, 2014 . Proceedings. LNCS, vol. 8765 . Springer ( 2014 )

16. Muller, G., Bergmann , R.: CookingCAKE: A framework for the adaptation of cooking recipes represented as work ows . In: Kendall-Morwick , J . (ed.) Workshop Proceedings of ICCBR 2015 , Frankfurt, Germany, 2015 . CEUR Workshop Proceedings , vol. 1520 . CEUR-WS.org ( 2015 )

17. Muller, G., Bergmann , R.: Generalization of work ows in process-oriented casebased reasoning . In: Russell, I. , Eberle , W. (eds.) Proceedings of FLAIRS 2015 , Hollywood, Florida. 2015 . AAAI Press ( 2015 )

18. Muller, G., Bergmann , R. : Learning and applying adaptation operators in processoriented case-based reasoning . In: Hullermeier, E., Minor , M. (eds.) Case-Based Reasoning Research and Development, ICCBR 2015 , Frankfurt am Main , Germany, 2015 , Proceedings. LNCS, vol. 9343 . Springer ( 2015 )

19. Muller, G., Bergmann , R.: POQL: A new query language for process-oriented casebased reasoning . In: Bergmann, R. , Gorg, S. , Muller, G. (eds.) Proceedings of the LWA 2015 Workshops , Trier, Germany, October 7- 9 , 2015 . CEUR Workshop Proceedings , vol. 1458 . CEUR-WS.org ( 2015 )

20. Muller, G., Bergmann , R. : Case completion of work ows for process-oriented casebased reasoning . In: Proceedings of ICCBR 2016 . Atlanta (Georgia) , USA ( 2016 )

21. Rittgen , P. : Quality and perceived usefulness of process models . In: Shin, S.Y. , Ossowski , S. , Schumacher , M. , Palakal , M.J. , Hung , C. (eds.) Proceedings of the ACM Symposium on Applied Computing (SAC) , Sierre, Switzerland,, 2010 . ACM ( 2010 )

22. Schumacher , P. : Work ow Extraction from Textual Process Descriptions . Ph.D. thesis , Wolfgang Goethe University, Frankfurt/Main ( 2016 )

23. Smyth , B. , Keane , M.T.: Retrieving adaptable cases: The role of adaptation knowledge in case retrieval . In: Wess, S. , Altho , K. , Richter , M.M. (eds.) Topics in Case-Based Reasoning , EWCBR- 93 , Kaiserslautern, Germany, 1993 . LNCS, vol. 837 . Springer ( 1993 )

24. Taylor , I.J., Deelman , E. , Gannon , D.B. , Shields , M. : Work ows for e-science - Scienti c work ows for grids . Springer, London ( 2007 )

25. Work ow Management Coalition: Work ow management coalition glossary & terminology . http://www.wfmc.org/docs/TC-1011_term_glossary_v3. pdf ( 1999 ), last access on 20-09-2016