Detection And Optimization Of The Process Model Based On Conformance Checking Approximation

The idea of process mining is to discover,monitor,and improve real processes by extracting knowledge from event logs available in today’s systems,and to establish relationships between real processes and their data and process models.Both consistency checking and model optimization are important process mining techniques.Conformance checking monitors the difference by comparing the behavior of process instances recorded in event logs with the process model,measuring the deviation between process execution and its normative or descriptive behavior,and ensuring that the discovered model or the model given by the system is consistent with the actual business process.Model optimization is committed to using some event log records in the actual process information to extend or optimize existing process model,because the may not want to revise the process model to adapt to the height of the unusual behavior,so can choose only optimization is often observed deviation behavior part,which is based on conformance checking deviation detected in as little as possible the changes to the original model at the same time so as to make the model more realistic.This paper mainly detects whether there is deviation between the process model and the actual business process based on approximate conformance,and then optimizes the process model by mining hidden transitions or configuring Petri net model.For the actual business process to be modeled,the process model was established based on Petri net and the knowledge of behavior profile,and the behavior relation optimization model was used in time when the changes of actual business process were observed.In view of the existing process model,the conformance between the actual process collected and the process model is checked based on the approximate conformance,and the noise and effective low-frequency trace are distinguished by the approximate conformance relaxedly,and then the process model is optimized based on the log trace after filtering noise.To sum up,the main contents are as follows:(1)Firstly,a Petri net model is established based on the behavior profile relation in view of the transfer from offline process to online process of IT service release of a company.Firstly,the activity sequences collected in offline processes are summarized as event log sequences,and the initial process model of the system is generated based on Petri net by analyzing the behavior profile relation between activities in the logs.When the actual business process changes are detected,the difference of behavior profile relationship is obtained by analyzing the changing online event log and the original offline event log,and the changing model block is determined based on the behavior relationship,and then the initial model is optimized by using the direct dependence of activities in the changing model block.By calculating the alignment values of the new event log on the line with the original model and the optimization model respectively,IT shows that the optimization model has a higher conformance with the actual process.(2)Secondly,the problem of mining hidden transitions based on approximate conformance results is studied,and the pre-processing method of filtering noise and the method based on hidden transition optimization model are proposed.The edit distance function is used to calculate the approximate fitness and upper and lower bounds of the log sequence and the initial model behavior subset.At the same time,by relaxing the interval definition of fitness,only the logs that are most likely to be noise are removed,and the low-frequency logs that conform to the normal flow are left.In order to find the direct input and output of each activity in the effective low frequency sequence,the bindings activity set is proposed,and then the reasonable hidden changes in the model are mined as a support.The feasibility of this method is verified by an example.(3)Finally,aiming at the Petri net model of direct transformation of combined flow diagrams of similar processes with partial same behaviors given by the organization or system,the method of using configuration optimization model based on the results of approximate conformance check is proposed.The simulation method is used to widen the fitness interval to filter noise.Different from traditional configuration,this method is modified in the original configuration concept.The possibility of direct blocking is cancelled and the model is temporarily closed when the configuration rule is triggered.Based on the behavior relationship mining configured points and corresponding configuration trigger rules,the configured model with configuration rules was obtained,and the method was illustrated based on an example of the admission process of newly admitted patients during the epidemic.Figure [13] Table [21] Reference [103]...