Subprocess Discovery

Many real world domains are characterized by low structured processes, whose execution is mainly driven by the decisions taken by the actors which perform the processes. Classical process analysis techniques, in particular process mining techniques, usually return poor outcomes for these processes. As a remedy, we propose a graph-based methodology of analysis, aimed at discovering the most relevant subprocesses shared among the set of heterogeneous process instances. We expect these subprocesses are able to provide useful insights about the process which cannot be derived by considering the whole process, thus resulting a valuable aid for process analysis.