Research And Application Of Merging Event Logs For Process Mining

Process mining is a hot research topic in the field of process management and data mining.The current process mining techniques and tools are based on single event log as input.But in an actual business environment,in most cases,these information systems do not manage the entire business process,but some activities in the process.Thus the recorded data of the process execution which supported by different computer systems is scattered into different log files.Therefore,it is necessary to merge the scattered data into one single log file when applying current process mining techniques and tools for entire process mining and analysis.This mission is still challenging,however,because of the difficulty of unifying the case’s identification of cross IT system and the complex relationships between cases in two logs and the possible lack of information for the merging.In order to solve the above problems,this paper studies and analyzes the problems that need to be solved in merging event logs for process mining,chooses the event log merging at case level,considers the artificial immune system with the immune memory bank as the foundation of log merging method,analyzes and designs the immune operators(or steps),and finally achieves a feasible event log merging algorithm.The work is as follows:1)Reviews the domestic and foreign related research and introduces the concept and definition of matching matrix,and then regards the event log merging for process mining as a kind of search and optimization problems based on the formal definition.2)A merging approach with a Hybrid Artificial Immune Algorithm(HAIA)is presented in order to achieve the log merging with one to many,many to one and many to many complex relationships between cases in the two event logs.In the merging approach,a matching matrix is an individual,a heuristic method is used to generate the initial population,and the clonal selection and mutation principles are selected as its underlying principle,which requires the matching process to undergo iterations of clonal selection,hyper mutation and receptor editing so as to get the best solution.3)Two factors,occurrence frequency and temporal relation,are designed in the affinity function to evaluate the individuals in the population.Respectively,it adopts the two dimensions,"quantity" and "time",to evaluate the evolution of the individual,makes the matching relationship between the cases more accurate.4)Immunological memory and simulated annealing are exploited to make the artificial immune merging jumping out from the trap of local optima.Finally,using a number of different data sets to conduct the experiment.The results show that the merging success rates are as high as 90% or more,prove that the validity of this log merging approach with many to many relationships between cases.By comparing the AIA algorithm of Claes et al,this paper’s approach can improve the efficiency in about 20%.Finally,compared with the random method to generate initial population,the heuristic method can accelerate the speed of immune evolution.