Abstract: To make event log which contained vast activities mine process model effectively,a decomposed mining method for process model based on behavior feature net was proposed.The behavior footprint between activities were found based on activity logs and achieve corresponding behavior matrix as well.The behavior relation graph of activities was computed with behavioral matrix,and the clustering of activities was generated.Sub-log was filtered by existing mining algorithm for discovering subnets,and interface places were added to each subnet to structure behavior feature nets of sub-logs.On the basis of behavior feature nets,the complete net was synthesized with viewpoint of synthetic nets,and so as to mine the process model.The effectiveness of the proposed method was verified by a simulation analysis.

Key words: process model mining, behavior footprint, behavior matrix, cluster, behavior feature net, synthesis

摘要： 为了使包含活动数目较多的事件日志有效挖掘流程模型,提出基于行为特征网的流程模型分解挖掘方法,基于活动日志确定各活动间的行为足迹关系,推得相应的行为矩阵;结合行为矩阵计算行为关系图,从而产生活动聚类;通过现存挖掘算法过滤子日志挖掘子网,并对子网添加接口库所形成子网行为特征网;在行为特征网的基础上,运用合成网的观点合成整网,以此挖掘流程模型。最后通过仿真分析验证了该分解挖掘方法的有效性。

关键词: 流程模型挖掘, 行为足迹, 行为矩阵, 聚类, 行为特征网, 合成