Business process graph matching based on vectorial signatures

During the last decade, business processes have gained in popularity leading to a rapid proliferation of business process model collections. The need for efficient and fast systems for similarity search or matching of process models from these collections is becoming an important issue with the increasing of the number of models. Graph-based approaches are very efficient for measuring the similarity of process models. However, they suffer from high complexity that grows exponentially with the size of the graphs. In this paper, we propose an approach for comparing and evaluating the similarity of process models based on vectorial signatures, taking into account the loops which may occur in the process. To reduce the complexity of matching, we split the process graphs into paths, then calculate their similarity based on the similarity of the different vectorial signatures of their paths. Experimental evaluations show the good performance of the proposed approach using several measures.