Abstract:

The existing process mining techniques and tools are on the basis of a single log file.In actual business process environment,however,a business process may be supported by different computer systems,so that actual process data will be recorded into multiple log files.Therefore,it is necessary to merge the multiple recorded data into one log file for further global process mining and analysis.In this paper,an automatic method is proposed to merge event logs by combining an artificial immune algorithm and simulated annealing.In the method,on the basis of the characteristics of the process logs of multiple IT systems,two operators,namely,the occurrence frequency of activity sequences and the time overlap area between mergeable cases,are taken into account in an affinity function,so as to improve the accuracy of matching cases and the practicality of the proposed method.Moreover,the simulated annealing selection is introduced into the evolution of populations so as to solve the problems of the premature and continuous degradation of artificial immune algorithm,and the immunological memory is also introduced to preserve the diversity of populations and avoid their local convergence.Experiment results show that the proposed method achieves a merging success rate of more than 90%,and it can ensure that process mining results are correct,and that,as compared with the traditional log data-merging method on the basis of artificial immunity,the proposed method speeds up convergence significantly and increases merging efficiency.

Key words: process mining, log data merging, artificial immune algorithm, simulated annealing