Faster and More Focused Control-Flow Analysis for Business Process Models through SESE Decomposition

We present a technique to enhance control-flow analysis of business process models. The technique considerably speeds up the analysis and improves the diagnostic information that is given to the user to fix control-flow errors. The technique consists of two parts: Firstly, the process model is decomposed into single-entry-single-exit (SESE) fragments, which are usually substantially smaller than the original process. This decomposition is done in linear time. Secondly, each fragment is analyzed in isolation using a fast heuristic that can analyze many of the fragments occurring in practice. Any remaining fragments that are not covered by the heuristic can then be analyzed using any known complete analysis technique. We used our technique in a case study with more than 340 real business processes modeled with the IBM WebSphere Business Modeler. The results suggest that control-flow analysis of many real process models is feasible without significant delay (less than a second). Therefore, control-flow analysis could be used frequently during editing time, which allows errors to be caught at earliest possible time.

A shorter version of this report has been published in volume 4749 of Springer Lecture Notes in Computer Science, entitled "Service-Oriented Computing – ICSOC 2007", the Proceedings of the Fifth International Conference on Service-Oriented Computing, Vienna, Austria, September 17-20, 2007, pp. 43-55. Copyright (c) Springer-Verlag, 2007. DOI: 10.1007/978-3-540-74974-5_4.