An improved structure-based approach to measure similarity of business process models

It is common for large enterprises or organizations to maintain repositories of process models. This paper focuses on how to measure the similarity and construct matching relations more effectively between process models. To resolve exponential time complexity to match node compositions, we proposed a graph-edit distance similarity metric based on SESE process fragments, and a greedy algorithm is employed to construct the optimal matching relations. Then a method to construct matching relations of fragments based on process structure tree is proposed. Finally, a comparative experiment based on real-world process models from BPM AI repository is conducted to evaluate the effectiveness and efficiency of our approach.