Methods And Applications Of Business Process Mining Based On Infrequent Behavior

Process mining is a hot topic in the field of process management,which can discover,monitor and improve business process through extracting information from event logs,building on process model-driven approaches and data mining.With the development of information,further research and analysis of process mining is needed to meet the increasing challenges.The main content of this paper is to analyze and study the multi-module system and infrequent behavior in the area of process mining.Nowadays,many existing research methods are used to deal with single module problems,which have certain limitations to multi-module systems In this paper a multi-module mining method was proposed based on feature net and modular net.Firstly,each modular log is separated from the event log according to the modular mark,and modular net is extracted from the modular log by Language—Based Regions.Then,a character net mining method based on the character communication is proposed by analyzing the interaction between modules.Finally,a method of mining the complete Petri net model is proposed by integrating feature nets and modular nets based on the property of open Petri nets.Besides,the research on the infrequent behavior in the log is not perfect.Usually,all infrequent behaviors were filted as noise.In the actual system some infrequent behaviors contains important system information which cannot be simply removed.For example,the escape behavior in space-craft system,the safety bag in vehicle system and so on.Although these behaviors rarely occur during the operation of the entire system,they are particularly important.In order to solve this problem,an infrequent behavior filtering method is proposed based on casual net.Firstly,a causal net is constructed based on causality between events in event logs.Based on causal net filtering method,the real noise behaviors are removed and the infrequent behaviors containing system information are preserved.Then,a Petri net model,which is more conducive to system analysis,is mined from the event log and causal net by the mining method proposed in this paper.Through the analysis and study of multi-module system and infrequent behavior,the method of mining multi-module system and the filtering method of infrequent behavior are put forward in this paper.The validity and consistency of the method are verified by the example and the method of consistency evaluation.