Repairing Process Models Containing Specific Structures Via Logic Petri Nets

With the continuous development of workflow technology,business process reengineering theory and traditional production process theory,business process management has been well applied in various enterprises and organizations.The log data generated in the process of business management can be recorded and processed in detail through information technology,which provides many valuable data information for enterprises and organizations.These logs can be used by enterprises and organizations to detect business processes and facilitate business process optimization.Process mining technology can extract process knowledge from the general event log of modern information system,which can establish process model.And we then detect and diagnose the deviation between process model and observation behavior through consistent diagnosis technology.The existing model repair method only focuses on the index of fitting degree,and the other metrics are less considered.Moreover,the repaired model has different parts describing the same behavior,which increases the complexity of the model.It does not consider the relationship between the new and the original activities,and the relationship between activities of concurrent structures.Based on this situation,we use the logical relationship between the trace activities in the event log and the model transitions,and the logical relationship between the concurrent transition selection structures in the event log.This paper proposes a repaired technique based on logic Petri nets.The repaired model can better avoid the existing method,due to the addition of self-loops,it leads to sub-processes repeatedly appeared in the process of repairing models.Combining Petri net with process tree,concurrency model in the leaf nodes based on the search process,can locate the concurrent structure of Petri net.Then according to the concurrent structures through logic repair,the models can be repaired better.For process models containing choice structures,the model repair approaches are proposed.Specifically,some relations between the transitions in the choice structure are studied in order to decide position where to repair the model.Then according to the choice structures through logic repair,the models can be repaired better.In view of the need to add new activities to construct and repair process models containing the loop return structure,we identify the deviations between the transitions in the models and the activities in the logs.Then,a method is designed to construct and repair the process models containing a loop return structure.According to the relations between the transitions in the process models,logic transitions in LPN can be constructed.Then according to the loop return structures through logic repair,the models can be repaired better.When the event logs and the corresponding process model are inconsistent during the consistency checking,i.e.,the values of quality metrics are low,we need to repair the process model.The difference between a process model and an actual operational process in event logs is identified by using a suitable deviation recognition approach.Models can be thus repaired by adding or deleting elements or operations at appropriate locations.Therefore,this paper proposes the logic Petri model repair method based on the three different structures,which can correct the complex actual process with corresponding structures and further improve the fitness and precision.Finally,according to the verification and comparative analysis of simulation experiments,the correctness and effectiveness of the logic Petri model repair method based on the three different structures are illustrated,and the fitting degree and accuracy of the correction model obtained are higher.