Modular Process Model Repair Method Based On Behavioral Profile

Model repair is a new process mining technology,which takes the event log and process model as input,finds the deviation in the process model by analyzing the log,and then repairs the model.At present,we can study the real situation of the system through the event log,and use the process mining technology to get the process model.In the field of big data,the amount of data generated by various systems is huge and complex,and new activities are often generated in the process operation,resulting in the original process model can not accurately reflect the actual process,so it is necessary to repair the process model.The existing process model repair methods do not describe the systematic repair process,and most of the repair process will traverse the event log countless times,which may lead to a situation that some models without deviation are repaired,resulting in unnecessary work.Moreover,the repair results of these repair methods only focus on one index of the evaluation model,such as the degree of fit,but ignore the accuracy.In order to solve this problem,we proposed a modular process model repair method based on behavior profile.Firstly,about the systematic repair method,this dissertation gives the basic process of the general model repair method based on the idea of divide and conquer,namely the modular process model repair method.The model repair is divided into five modules:decomposition block,filter block,selection block,repair block and combination block,and gives the corresponding algorithms of filter module and selection module.The initial process model is decomposed into several model fragments by using the maximum decomposition technology,and the superior sub model is filtered out.The filtered superior sub model does not need to be traversed for the repair process.The inferior sub model that needs to be repaired is selected by selecting the module algorithm for repair.Finally,the repaired model fragments and superior sub model are combined to complete the systematic repair process of the model.Secondly,aiming at the problem of high fitting degree of the repaired model without paying attention to the accuracy,this paper introduces the reachable identifier of Petri nets according to the modeling repair method,finds the deviation in the process and repairs it through the optimal extension alignment rule.In the existing methods,the process model fragment is directly added to the original process model as a self loop,allowing this sub process to occur repeatedly,which will lead to the low accuracy of the repaired model.This method avoids the emergence of self loop and greatly improves the accuracy of the process model.Finally,the online shopping return and exchange process model have been optimized and repaired according to the above methods,and the optimized model can well reflect the actual business,and the simulation results approved that the method is feasible and effective.Figure[31]Table[10]Reference[83]...