期刊
EXPERT SYSTEMS WITH APPLICATIONS
卷 174, 期 -, 页码 -出版社
PERGAMON-ELSEVIER SCIENCE LTD
DOI: 10.1016/j.eswa.2021.114662
关键词
Process mining; Model discovery; Heuristic rules; Event logs; Filtering; Outlier behaviour
This paper presents an approach to discover business process models based on BPMN standard from event logs. Experimental results show that the process models derived by this method exhibit better results in fitness and precision.
Information systems daily register a large amount of data in event logs. All these data can be used for the organizations to automatically discover their process models. However, the automatic construction of simple process models with consistently high and balanced fitness and precision remains a challenging task, which has attracted the attention in the scientific and organizational communities to develop suitable methods for creating high-quality business process models. In this paper, we present an approach to discover the model of a business process based on the Business Process Model and Notation (BPMN) standard from its behavior observed in event logs. Particularly, we propose: (1) a method to detect outlier behavior in a given event log; (2) a set of heuristic rules to discover join gateways associated with the closing of each of the split gateways in conformance with the rules of BPMN models; and (3) P-Miner, a tool that realizes our proposed approach to automatically discover and visualize process models. We use a state of the art process discovery technique as a basis for discovering XOR/ AND gateways in the models. A set of experiments was carried out on real and artificial event logs for evaluating our proposal, considering the fitness and precision metrics to determine the quality of the built models. The amount of gateways in the resulting BPMN model and the time required to build the process models were estimated. The results reveal that the process models derived by our proposed approach exhibited competitive and more balanced results in fitness and precision than those derived with other discovery techniques.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据