Discovering models of parallel workflow processes from incomplete event logs

Abstract

α-algorithm is able to discover a large class of workflow (WF) nets based on the behavior recorded in event logs, with the main limiting assumption that the event log is complete. Our research has been aimed at finding ways of business process models discovering based on examples of traces, i.e., logs of workflow actions that do not meet the requirement of completeness. In this aim, we have modified the existing and introduced a new relation between activities recorded in the event log, which has led to a partial correction of the process models discovering techniques, including the α-algorithm. We have also introduced the notions of causally and weakly complete logs, from which our modified algorithm can produce the same result as the original algorithm from complete logs. The effect of these modifications on the speed of the process model discovering is mostly evident for business processes in which many activities can be performed in parallel. Therefore, this paper presents preliminary results obtained from the investigation of opportunities to discover models of parallel processes based on incomplete event logs.