Pictures copyrighted

A.J.M.M. Weijters - Flexible Heuristics Miner (FHM)

Modern enterprises are increasingly becoming dependent on the quality of their business processes. A necessary first step to improve a business process is a correct understanding of this process. Process mining aims at the extraction of non-trivial information from running business process data sets (i.e., event logs or transition logs) and can contribute to this understanding. Event logs are the starting point for process mining techniques. Control-flow mining, conformance checking or performance analyses are possible applications. The main focus of the research presented in this paper is on control-flow mining, i.e., the induction of non-trivial process information from running business process expressed in a process model. Control-flow mining is a necessary starting point for many other process mining techniques.

A lot of work in this sub domain is already done. Most of early solutions try to model all the recorded behavior in the event log by using a formal process modeling language (e.g. the Petri net formalism). However these kinds of approaches run in problems in less structured domains such as the ones that can be seen in health care. The resulting models may easily become unreadable if the model contains a high number of activities and complex relationships. An important motivation for the approach as presented in this presentation is the development of a flexible control-flow mining algorithm that performs well in practical situation and with results that are easily to understand.