Abstract: Token replay is one of the most widely used techniques in conformance checking that enables cost-effective computation of the fitness between process models and event logs.However,it is unable to effectively dealing with duplicate tasks.When encountering duplicate tasks,the Token replay method activates different tasks with the same tag name,thereby affecting the subsequent replay path of the trace and consequently influencing the evaluation results of the process model.Although alignment-based fitness evaluation methods can effectively handle duplicate tasks the process of finding the optimal alignment is extremely time-consuming and not suitable for large-scale event logs.To address these issues and to improve the Token replay process,a fitness evaluation method for duplicate tasks called Fitness Evaluation Approach for Duplicate tasks in process (FEAD) was proposed,which could effective handle duplicate tasks.All possible paths when replaying duplicate tasks were considered to select the optimal path.Experimental comparisons were conducted on multiple artificially constructed datasets and real datasets.The results demonstrated that the proposed FEAD method could effectively handle duplicate tasks and had significantly lower time consumption compared to alignment-based fitness evaluation methods.

Key words: process mining, conformance checking, Petri nets, duplicate tasks

摘要： 基于托肯重放的适应性评估方法是流程挖掘的一致性检查中应用最广泛的方法之一,它能够以较低的时间代价计算出流程模型和事件日志之间的适应性,但无法有效处理重复任务。在遇到重复任务时,该方法会激活具有相同标签名的不同任务,这将影响轨迹后续的重放路径,从而对流程模型的评估结果产生影响。基于对齐的适应性评估方法虽然能有效处理重复任务,但是寻找最优对齐的过程极其耗时,无法应用于大型事件日志。该研究旨在解决上述问题,改进托肯重放的过程,提出了一种可处理流程模型中重复任务的适应性评估方法（FEAD）,FEAD可以有效处理重复任务,并考虑重放重复任务时可能执行的所有路径,从中选出最优。通过在多个人工构建的数据集和真实数据集上进行实验,结果表明FEAD可以有效处理重复任务,且在时间消耗上远远低于基于对齐的适应性评估方法。

关键词: 流程挖掘, 一致性检查, Petri网, 重复任务