Research On Methed Of Process Mining For Non Free Choice Construct

Process mining is a technique for mining information from event logs recorded by information systems to analyze,discover,improve,and manage business processes,and to detect bottlenecks in business processes.In process mining,the main task is to build a process model that can represent the actual business operation.A process model represents the dependencies between activities in the process.This thesis mainly studies the generation of models from the perspective of control flow,focusing on the order relationship between activities.It is always a challenge in the field of process mining to dig out the non free choice constructs of synchronization and selection from the event log.Most algorithms can't detect the concurrency relationship between activities better.In order to solve the problem that the mining result deviates from the actual business process,the related mining method of the non free choice construct itself and the existing combined genetic algorithm.In this thesis,the traditional genetic process mining method is improved.According to the non free choice construct to be explored,the order definition that can judge the long distance dependence is added,and the initial population is more biased,which is prepared for the subsequent evolutionary calculation.Adequate quality individuals,thus narrowing the search space and avoiding the existence of inferior individuals with less correlation with log behavior.Evaluation of the resulting model using three general indicators of fitness,precision,and generalization.The results show that the excavated process model can reflect the actual business process very well.The main research contents of this thesis are as follows:Firstly,the composition and characteristics of the event log and the representation of the process model Petri net and causal matrix were studied.The event log is the starting point of the whole process mining.It analyzes the complex structures such as loops,parallels,non free choices and hidden tasks.It is possible to mine these complex structures from the event log.It is important to verify the correctness of the results.Secondly,a comparative study of current process mining technologies is conducted in this research.The genetic process mining,? algorithm and heuristic process mining method are studied in detail.Thirdly,an improved genetic process mining algorithm is proposed,which can discover the non free choice construct.Reduce search space by building quality initial populations.The feasibility of this method is verified by experiments.So the excavated process model can reflect the real business process very well.