Research On Repair And Optimization Of Business Process Model Based On Petri Net

Currently,process models can be mined from a large number of event logs and remagnified from most of them.However,a few logs that deviate from the process model are also valid,and model repair is a good method to make the event log fit better with the process model.Model repair technology is widely used in many fields.Model repair can not only achieve the purpose of completely replaying the log,but also keep the similarity with the original model to the greatest extent and retain the value of the original model.Therefore,more and more scholars focus on the study of model repair.Based on the Petri Net and the knowledge related to the behavior profile,this paper builds a process model for a given event log according to the behavior profile relationship between activity events.Then,through the consistency check of the process model,problems can be found and analyzed,and the ideal model effect can be achieved by means of model repair or model optimization.The main research contents of this dissertation include:(1)Firstly,the problem of continuous log movement is further studied and analyzed.Many scholars have studied the problem of process model repair with log movement.one of the most classic repair methods of existing methods for insert-skip activity repair method,this method is easy to understand and use for a single log mobile.However,in most cases you will see continuous log movements,so the insertion of multiple active events in the same location can be cumbersome to the model.Therefore,based on the existing subprocess repair algorithms,an effective algorithm to find the set of repairable log movement place for the process model repair problem with continuous log movement is proposed to assist model repair.From the point of view of cost alignment,model repair can be implemented effectively according to the detected log movement part with this algorithm.(2)Secondly,to solve the problem of process model repair with concurrent events,the methods of screening concurrent e vents and reconstructing subprocess model repair are proposed.In the existing model repair literature,most of the analysis and research are directly based on the part of asynchronous movement,thus neglecting the potential relationship between asynchronous movement and adjacent synchronous movement.Therefore,this paper first finds out the optimal alignment between the event log and the process model.Based on this,aiming at the potential concurrent events between the log movement and the adjacent synchronous movement,a method of screening concurrent events is proposed.This method can effectively screen out the concurrent events for repair.Aiming at the process model with such concurrent events,this paper proposes a repair analysis method for the process model of concurrent events based on Petri nets.After finding concurrency events that meet the conditions,the method reconstructs a subprocess of concurrency structure according to the relevant knowledge of concurrency relation of behavior contour,so that the deviation between the event log and the process model is zero,and the process model can better reflect the actual event log.(3)Finally,in view of the problem of multiple process model interaction and merger,the process model is repaired and optimized by consistency check.Most of the existing literature studies the interaction and merging of related process models from different roles.However,in fact,there are many related process models from the same role.For example,starting from the role of online shoppers,different shopping policies will lead to different process models in the process of group shopping,so there will be more than one process model.For the problem of multiple process models starting from the same role,the implementation in the same system is scattered and not centralized,which is not conducive to the discovery and analysis of problems,so it is necessary to merge the related process models.Simply merging them may not achieve ideal results.Therefore,this paper combines such problems and realizes further repair and optimization based on relevant knowledge of Petri net and concepts related to behavior profile.Finally,the effectiveness of the optimized model is verified by an example of the role of online shoppers.Figure[27]table[8]reference[104]...