Petri Nets Based Recognition of Model Deviation Domains and Model Repair

Process mining techniques can be used to discover process models from event logs. Event logs and process model can be contrasted by conformance checking techniques. And conformance checking techniques can be used to detect the deviations between observed behaviors and process model. However, existing techniques of process mining concern with discovering these deviations, but not support to repair the process model easily and make the process model more related to the real process. So in this paper we propose a token-based identification method of model deviation domains and a token-based technique of model repair (static model repair and dynamic model repair) through techniques of conformance checking and dynamic behaviors of workflow net. Model deviation domains can be identified effectively though the flow direction of token. We can repair process model according to model deviation domains. And we also can repair the real complex process accurately which has the structures of complicated circulation and choice. In this paper, the effectiveness and the correctness of techniques are illustrated through contrast experiment and analysis with other techniques.