Process Mining: Algorithm for S-Coverable Workflow Nets

• Authors: Jianchun She;Dongqing Yang
• Venue: WKDD '09 Proceedings of the 2009 Second International Workshop on Knowledge Discovery and Data Mining
• Year: 2009

To discover process models from event logs has recently aroused many researchers’ interest in the area of process mining. Notwithstanding the interest and related efforts, existing algorithms are far from being satisfactory. For example, some researchers have proved that α-algorithm is capable of discovering the processes of the so-called SWF-nets without short loops; however, α-algorithm has been found to contain some severe limitations. This paper is therefore to introduce the notation of S-Coverable workflow nets, to reach theorems about the characteristics of Sound S-Coverable Workflow Nets, and to develop a new process mining algorithm, namely, algorithm S. On such basis, suggested in the paper is a new approach to dealing with the problem of hidden transition discovering, an approach that, by means of the pretreatment of such hidden tasks, allows algorithm S to discover process models that will help preserve better structures. Theorems thus reached are applicable not only to process mining, but also to process modeling and process model checking.