ModelMatching.PDF

Probabilistic Optimization of Semantic Process Model Matching Henrik Leopold1, Mathias Niepert2, Matthias Weidlich3, Jan Mendling4, Remco Dijkman5, and Heiner Stuckenschmidt2 1 Humboldt-Universit¨atzu Berlin, Unter den Linden 6, 10099 Berlin, Germany henrik.leopold@wiwi.hu-berlin.de 2 Universit¨atMannheim, 68159 Mannheim, Germany mathias|heiner@informatik.uni-mannheim.de 3 Technion - Israel Institute of Technology, Technion City, 32000 Haifa, Israel weidlich@tx.technion.ac.il 4 Wirtschaftsuniversit¨atWien, Augasse 2-6, 1090 Vienna, Austria jan.mendling@wu.ac.at 5 Eindhoven University of Technology, 5600 MB Eindhoven, The Netherlands r.m.dijkman@tue.nl Abstract. Business process models are increasingly used by companies, often yielding repositories of several thousand models. These models are of great value for business analysis such as service identi?cation or process standardization. A problem is though that many of these analyses require the pairwise comparison of process models, which is hardly feasible to do manually given an extensive number of models. While the computation of similarity between a pair of process models has been intensively studied in recent years, there is a notable gap on automatically matching activities of two process models. In this paper, we develop an approach based on semantic techniques and probabilistic optimization. We evaluate our approach using a sample of admission processes from di?erent universities. 1 Introduction Business process models are increasingly used by companies for documentation purposes. A process documentation initiative stores an extensive amount of process models in a centralized process repository. This amount can easily rise to