An Approach for Repairing Process Models Based on Logic Petri Nets

Process mining technology can extract process knowledge from event logs that are generated from information systems. It can construct a process model by mining the event logs. The model usually needs to be repaired to accurately describe the business process realized by the information system. The existing methods for repairing process models cannot enhance efficiently some model consistency metrics, such as the fitness, precision, and simplicity. Thus, a new model repair approach is proposed based on an extended Petri net named logic Petri net in this paper. It can improve the model's fitness and precision comparing with the existing work. First, it builds process models via logic Petri nets. Next, approaches are proposed to repair the process models containing a causal relation and a concurrent relation, respectively. Specifically, a precursor set and a successor set of activities are defined and the relation of the elements in each of them is determined. Finally, we give some cases related to a thoracic surgery process in a hospital and conduct experiments to illustrate the correctness and effectiveness of the proposed approach.