Abstract: The purpose of process discovery is to construct a process model based on business process execution data recorded in an event log.Many situations may cause hidden transitions to appear in the process model,while the execution of hidden transitions does not appear in event logs.Therefore,mining of hidden transitions has been one of the difficult problems of process mining.Existing methods of hidden transition mining have some limitations when discovering the hidden transition in concurrent structures,and may cause some redundant hidden transitions.Based on these,a new method for mining process models with hidden transitions was proposed.The basic behavior relationship between activities was introduced based on the event log,the hidden transitions of and-gateway type and loop type were discovered using concurrency interleaving order relation and loop interleaving order relation respectively.According to the minimum and maximum behavior distance between activities in the event log,pairs of activities were found that a hidden transition of skip type may appear between them.The minimum behavior distance of these activities pairs based on the model and concurrent structure was further analyzed to determine hidden transitions of skip type,and constantly optimize the initial model.The process model with multiple types of hidden transitions was obtained.Experimental results showed that the proposed method could find multiple types of hidden transitions correctly,and the number of redundant hidden transitions was much less than the existing methods.Meanwhile,it effectively improved the fitness of the model without reducing the precision of the model.

Key words: labeled Petri net, hidden transition, behavioral distance, process mining, behavioral relationship

摘要： 过程发现的目的是基于记录在事件日志中的业务过程的执行数据发现过程模型,由于一些原因导致过程模型中可能会出现隐变迁,而这些隐变迁的执行又不出现在事件日志中,因此隐变迁的挖掘是过程挖掘的难点之一。已有隐变迁挖掘方法对解决并发结构中的隐变迁存在不足,且可能出现一些冗余的隐变迁。基于此,提出一种带隐变迁的过程模型挖掘新方法,首先基于日志分析活动的基本行为关系,通过并发交叉关系和循环交叉关系来发现and网关类型和循环类型的隐变迁。然后,根据活动基于日志的最小和最大行为距离寻找可能存在skip类型隐变迁的活动对,进一步分析该活动对基于模型和并发结构的最小行为距离,以发现skip类型的隐变迁,并不断优化初始模型,最终得到带多种类型隐变迁的过程模型。实验结果表明,该方法能正确地发现多类型隐变迁,相对现有隐变迁挖掘方法,所提方法能显著降低模型中冗余隐变迁的个数,同时在不降低模型精确度的前提下,有效地改善了模型的适合度。

关键词: 标签Petri网, 隐变迁, 行为距离, 过程挖掘, 行为关系