Enabling reuse of process models through the detection of similar process parts

Abstract

Many companies use business process modeling to support various improvements initiatives leading to an increasing number of process models. Typically, these models are stored in a collection containing several hundreds of process models. In many cases, process models are overlapping, although parts could be easily reused saving costs and efforts. Different labeling styles and evolving process models complicate the detection of reusable model parts. In this paper, we propose a novel approach for the detection of equivalent and similar process model parts that exploits semantic comparison of activity labels and behavioral comparison of control flow. We evaluate our approach on the SAP Reference Model, a collection with 604 process models. The evaluation reveals insights for the thresholds of semantic and behavioral similarity of process models as well as their influence for similar process part detection. Hence, we identify five candidate groups with specific similarity properties that contain reoccurring process parts. © 2013 Springer-Verlag Berlin Heidelberg.