Abstract: The business process always suffers from notorious problems, such as data race and deadlock. Moreover, these problems are extremely difficult to track down by testing and debugging. This paper proposes a novel detection approach for the BPEL process, which is based on the graph theory. The approach first transforms the BPEL process into a graphic model-BSG (BPEL Segment Graph). Then, the deadlocks are detected by finding strongly connected components of BSG. Meanwhile, data races are detected by determining the partial order among nodes in BSG, which in turn can be attained by deciding the reachability among them. It also employs the technique of constraint solving to enhance the accuracy of the detection result. Data races and deadlocks can be effectively detected by this approach, as demonstrated by practical application, thus enhancing the reliability of BPEL processes.

Key words: business process execution language(BPEL), BPEL segment graph(BSG), data race detection, deadlock detection, constraint solving