Translating Message Sequence Charts to other Process Languages using Process Mining

Abstract. Message Sequence Charts (MSCs) are a well known language for specifying scenarios that describe how dierent actors (e.g., system components, people, or organizations) interact. MSCs are often used as a starting point for software analysts to discuss the behavior of a system with dierent stakeholders. Often such discussions lead to more com-plete behavioral models described by e.g. Event-driven Process Chains (EPCs), UML activity diagrams, BPMN models, Petri nets, etc. The contribution of this paper is to present a method that uses process min-ing to translate a set of MSCs that represent example scenarios into a complete process model, e.g., represented in terms of EPCs or Petri nets. Our approach takes MSCs and translates them into a special kind event logs. Unlike all known process mining techniques, we use a new approach that uses event logs containing explicit causal dependencies. This allows us to discover high-quality process models. The approach has been implemented in the process mining framework ProM.