Conformance checking for BPMN-based process models

Abstract

Measuring how well business process models conform to the execution of the process in reality is an important topic with many applications. While current conformance checking approaches are tailored to formal models such as Petri nets they lack support for domain-specific standards such as BPMN. In this paper we present two approaches for directly measuring the conformance of business process models based on BPMN elements and event logs. We define methods for extracting properties from such models that enable an easy comparison to event logs on a local level (i.e. for individual parts of the process and individual events). Furthermore, we present a method for replaying whole event logs on such models, allowing for a global conformance measure (i.e. on trace level). By utilising the previously extracted properties, we eliminate the need for expensive state-space exploration.