Conformance Checking And Model Repair Based On Logic Petri Nets

Process mining techniques can extract process knowledge from event logs commonly generated by modern information systems,so as to complete the process of monitoring,analysis and improvement,to achieve the purpose of enhancing business processes.Enterprises and organizations can detect business process and get a lot of information from event logs.Conformance checking is an important part of process mining research.The existing conformance checking methods are not sufficient to find the correct deviation location.Meanwhile,existing model repaired methods only focus on the fitness and the metrics of precision consider less.Instead,the repaired model exists situation that different parts describe the same behavior which increasing the complexity of models.And the methods without considering the relationship between new activities and the original activities of event logs.For common business process involved in the case of model repair,A conformance checking method based on extended footprint matrix is proposed.By comparing two extended footprint matrices,a precise deviation position between the log and the model is obtained.Relationships between transitions in the original model are incorrect,in this case repairing Petri nets are confused,and conflicts will arise between transitions.Based on this,a non loop model repair algorithm based on logic Petri nets is proposed.The method is simple and accurate to express the logical relation between the activities,and improves the fitting degree and accuracy of the modified results.In view of the complex loop structure in model repaired,a conformance checking method based on optimal alignment is proposed.we can accurately find the loop sub-log according to the continuous log move and the accurate positioning deviation of the place where the identification is located.Based on this,a loop model repair algorithm based on logic Petri nets is proposed.The technique is especially possible to better positioning to the starting and ending position of the fixed loop structure,and mining the logic function of the logic of transitions.Finally,the validity and correctness of this method are illustrated by simulation experiment and comparison analysis,and the fitting degree and accuracy are higher.