摘要： 工作流模式挖掘是数据挖掘新的研究领域，可以从工作流执行所产生的记录中还原工作流模式，能有效地应用于业务需求建模和业务流程重构等方面。该文提出一种基于偏序代数运算和三角优化规则的工作流模式挖掘算法，只需要对业务执行记录进行一次性扫描，就能在线性时间内识别并还原出工作流模式。研究实验表明，该算法能获得较优的工作流模式，完整性较高，能包含原工作流，受执行记录覆盖率影响不大。而执行记录覆盖率越高，该算法所得工作流模式与原工作流模式的一致性也越大，因此，算法具有较高的应用价值。

关键词: 工作流模式, 数据挖掘, 执行记录

Abstract: Workflow pattern mining is a new research area in data mining. Workflow schema which is effectively applied to business requirement modeling and business process reconstruction can be mined from certain execution logs. A method of Logs-Based Workflow Mining(LBWM) is proposed to implement workflow schema mining based on algebra of partial order and triangular optimization rules. Experiments show that scanning the execution logs once and good workflow pattern can be obtained in linear time. The result includes the original workflow and has high completeness, while the coverage rate of the logs is not the main influencing factor. The result gets higher consistency with the original workflow pattern when the coverage rate of the logs is higher.

Key words: workflow pattern, data mining, execution logs

中图分类号: 

• TP301.6