Conformance checking techniques can be used to diagnose differences between observed behavior and modeled behavior. Although these techniques can be used to measure the degree of conformance of a running service based on recorded event data (e.g., messages or transaction logs) and its specification, their application may produce “false negatives” because a private view (i.e., an implementation) that accords with its specification may deviate significantly. The implementation may reorder some activities without introducing any problems, yet traditional conformance checking would penalize such changes unjustifiably. To overcome this problem, we present a novel approach that determines a best matching private view. We show that among the infinitely many accordant private views, there is a canonical best matching private view. Although the current implementation and experiments are limited to acyclic service models, the approach can also be applied to cyclic service models.
I gave this talk at the 9th International Workshop on Web Services and Formal Methods, WS-FM 2012, September 6-7, 2012, Tallinn, Estonia.