University of Portsmouth
Models of business processes can easily become large and difficult to understand. Abstraction has proven to be an effective means to present a readable, high-level view of a business process model, by showing aggregated activities and leaving out irrelevant details. Yet, it is an open question how to combine activities into high-level tasks in a way that corresponds to such actions by experienced modelers. In this paper, an approach is presented that exploits semantic information within a process model, beyond structural information, to decide on which activities belong to one another.