Implementation and Practical Relevance of Method for Synthesis of Structured Process Models from Transition Systems Based on Hierarchy of Events

Process mining techniques can be used to analyze behavior of systems. It is well known that Petri nets can be considered as a widespread formalism to model systems. If transitions are labeled with symbols from a given alphabet, they are to be interpreted as an occurrence of events in the system. It should be noted that a Petri net can be derived from a finite transition system with the help of region-based Petri net synthesis algorithm. Usually a transition system is mined from an event log, which represents possible sequences of processes. The problem is that region-based synthesis algorithm generates precise models perfectly repeating the log, but at the same time these models don’t have clear structure and are not general. The idea of the proposed approach is to improve the synthesis algorithm in such a way, that according to some predefined knowledge on a process structure, redundant (implicit) places will not be generated. In the context of this work a method for synthesis of structured process models from transition systems based on hierarchy of events is offered. This topic is extremely relevant for the following reason. A hierarchy of events allows removing redundant places, which is impossible when using region-based synthesis algorithm only. It is crucial to have quality dimension called fitness remained equaled to one (like when applying region-based synthesis algorithm) as generated model should accurately reproduce the cases recorded in the log. Moreover, it is preferable to improve simplicity of the model comparing the result of classical algorithm. The tradeoff is precession as additional behavior is going to be allowed though. A proposed approach is also implemented via plug-in in java for Process mining framework (ProM) environment. Key words: process mining, petri net, event log, transition system, region, petri net synthesis, structured process model, ProM