Automatic pattern mining in repositories of graph-based process models

Philip Hake, Peter Fettke, Peter Loos

In: Tagungsband Multikonferenz Wirtschaftsinformatik. Multikonferenz Wirtschaftsinformatik (MKWI-16) March 9-11 Illmenau Germany Pages 1143-1154 2 Technische Universität Ilmenau 2016.


Due to increasing numbers of business process models developed by different modelers for numerous domains, the need for reducing complexity of models and the repositories they are contained in emerges. Hence, the approach at hand aims at identifying patterns in repositories of business process models. In contrast to approaches addressing the detection of business process model clones or similar business process models, we focus on identifying and extracting patterns in different business process models. Using the Refined Process Structure Tree (RPST), we derive trees of nested process fragments from different models and then determine patterns among these nested fragments. We therefore provide an overview of existing types of process fragments, which are adapted in other approaches exploiting the RPST, as well as criteria to detect process structure patterns based on these fragment types. Furthermore, we implement and evaluate our approach in the context of a large business process model corpus in order to give insights into recurring patterns used in business process modeling.

German Research Center for Artificial Intelligence
Deutsches Forschungszentrum für Künstliche Intelligenz