Workflow Clustering Method Based on Process Similarity
Process-centric information systems have been accumulating a mount of process models. Process designers continue to create new process models and they long for process analysis tools in various viewpoints. This paper proposes a novel approach of process analysis. Workflow clustering facilitates to analyze accumulated workflow process models and classify them into characteristic groups. The framework consists of two phases: domain classification and pattern analysis. Domain classification exploits an activity similarity measure, while pattern analysis does a transition similarity measure. Process models are represented as weighted complete dependency graphs, and then similarities among their graph vectors are estimated in consideration of relative frequency of each activity and transition. Finally, the models are clustered based on the similarities by a hierarchical clustering algorithm. We implemented the methodology and experimented sets of synthetic processes. Workflow clustering is adaptable to various process analyses, such as workflow recommendation, workflow mining, and process patterns analysis.
KeywordsActivity Birth Rate Business Process Management Transition Vector Activity Vector Cosine Measure
Unable to display preview. Download preview PDF.
- Cardoso, J.: How to Measure the Control-flow Complexity of Web processes and Workflows. In: Fischer, L. (ed.) Workflow Handbook 2005, WfMC, Lighthouse Point, pp. 199–212 (2005)Google Scholar
- Malone, T.W., Crowston, K., Herman, G.A.: Organizing Business Knowledge: The MIT Process Handbook. The MIT Press, Cambridge (2003)Google Scholar
- Zamir, O., Etzioni, O.: Web Document Clustering: A Feasibility Demonstration. In: Proc. 21th Int. ACM SIGIR Conference, pp. 46–54 (1998)Google Scholar