Abstract: Process mining aims to restore business process models from event logs to provide technical support for understanding,improving and optimizing business processes.Traditional process mining techniques focus on a single business object,which cannot reflect the complex association relationships such as one-to-many and many-to-many in real-life businesses.Therefore,the complex association relationship in the original data will be flattened,resulting in the lack of collaboration and interaction between multiple objects in the mining results.To solve the above problems,an object-centric process model discovery technique was proposed to support discovery of multi-object Petri net process models from the object-centric event logs.Furthermore,aiming at the correctness of multi-object Petri net model,a correctness verification method based on extended multi-object Petri net reachability analysis was proposed.Finally,the correctness and effectiveness of the proposed method were verified by comparing the application cases with the traditional process mining techniques.Process mining aims to restore business process models from event logs to provide technical support for understanding,improving and optimizing business processes.Traditional process mining techniques focus on a single business object,which cannot reflect the complex association relationships such as one-to-many and many-to-many in real-life businesses.Therefore,the complex association relationship in the original data will be flattened,resulting in the lack of collaboration and interaction between multiple objects in the mining results.To solve the above problems,an object-centric process model discovery technique was proposed to support discovery of multi-object Petri net process models from the object-centric event logs.Furthermore,aiming at the correctness of multi-object Petri net model,a correctness verification method based on extended multi-object Petri net reachability analysis was proposed.Finally,the correctness and effectiveness of the proposed method were verified by comparing the application cases with the traditional process mining techniques.

Key words: object-centric process mining, model discovery, multi-object Petri net, correctness verification

摘要： 流程挖掘旨在从企业信息系统的事件日志中还原业务流程模型,为理解、改进和优化业务流程提供技术支持。传统流程挖掘技术面向单一对象,无法体现实际流程对象中存在的一对多和多对多等复杂关联关系,若将传统流程挖掘技术直接应用到多对象业务流程中,需要将原始数据中的复杂关联关系扁平化,导致挖掘结果中多对象间的协同与交互关系缺失。针对以上问题,首先提出一种对象为中心流程模型挖掘方法,可以从对象为中心事件日志中挖掘多对象Petri网流程模型,来精准描述多对象复杂交互业务行为。进而,针对多对象Petri网模型的正确性验证问题,提出了基于扩展多对象Petri网可达性分析的正确性验证方法。最后,通过应用案例与传统流程挖掘方法比较验证了所提方法的正确性和有效性。

关键词: 对象为中心流程挖掘, 模型挖掘, 多对象Petri网, 正确性验证