Formal Modeling and Discovery of Hierarchical Business Processes: A Petri Net-Based Approach
本文定义了层次化Petri网（HPN）的概念，并提出一种从带有生命周期信息的事件日志中发现HPN的方法，同时给出将HPN转换为经典Petri网以应用现有质量度量的细节。实验表明该方法能有效发现层次化过程模型，并在模型质量上优于现有方法。
Business processes are critical for information systems to control workflows and deliver services. Although existing process discovery techniques can generate flat process models from business event logs, few of them have investigated the notion of hierarchy (i.e., subprocesses) yet. To fill the gap, this article first defines the concept of hierarchical Petri nets (HPNs), which can support the formal modeling and correctness verification of processes with subprocesses. Followed by that, we propose an approach which can effectively discover HPNs from event logs with lifecycle information. Moreover, to quantify the quality of discovered HPNs, details on how to transform an HPN to a classical Petri net are given such that existing metrics can be applied. All proposed approaches have been fully implemented in ProM, and experiments over both synthetic and real-life event logs demonstrate that our approach can effectively discover hierarchical process models. Specifically, compared to exiting approaches on processes discovery, our approach can generally perform better in terms of model quality.