A probabilistic method for business process verification: Reachability, Liveness and deadlock detection

In this paper we deal with business process verification for reachability, liveness and deadlock-freeness properties. Actual business process verification techniques use formal methods such as model checking that checks exhaustively temporal logics properties, however, in this case, the business process model must be mapped to an abstract mathematical model (like Petri nets, timed automata or process algebra) on which the verification process could be done. The fact that the verification is done on this second representation can fake the results if the transformation is either erroneous or over-constrained as well as if the proprieties are misspecified. In this article we illustrate that probability calculation can be used in business process verification by evaluating a disjunctive normal form of its conditions, moreover, the verification process could be done directly on the business process model and can handle all its functional elements. We have developed an application that numerically shows the performance of our proposed technique by verifying several workflows with various “known” flaws.