Process Matching: A Structural Approach for Business Process Search

The evaluation of "closeness" between services is always crucial in the process of Web service integration and reuse. However, matching services solely based on their inputs/outputs is insufficient for developers to understand their behavioral properties or other constraints. To go beyond this limitation, services should be matched and integrated at the process level. But existing approaches for business process matching mostly deploy symmetric measure which hardly reflect the perceptions of the requestor, i.e. some behaviors of the process may be highly desirable. To address these issues, this paper presents a structural approach to model Web service processes, and proposes an asymmetric way of computing process similarity based on edit distance. Process structure weights may be specified and contribute to the similarity value. In addition, to ensure a scalable and efficient search, two indexing schemes for the process model are developed.