Abstract: Hide transition exists in the business process, but it isn’t recorded in the log. Mining hide transition can restore the model and improve the running efficiency of the process. The existing methods are based on the direct dependency relationship between logs to mine hide transition, and do not consider their indirect dependencies, which have certain limitations. This paper proposes a method based on indirect dependencies mining hide transition. An initial model is constructed based on the behavior profile between activities in the event log, log sequence between the constraints of the body is found through the proposed indirect dependency table. Integral linear programming method is used to find the required quasi indirect relationship transition pairs, so as to explore the hide transition of quasi indirect relationship transition pairs. The hide transition is integrated into the initial model and the target model with hide transition is obtained. The validity of the method is verified by the concrete example analysis.

Key words: hide transition, process mining, quasi indirect dependence, integral linear programming, Petri net

摘要： 隐变迁存在于业务流程中，但在日志中未被记录，挖掘隐变迁能够还原模型并提高流程的运行效率。已有方法都是基于日志间直接依赖关系挖掘隐变迁，未考虑其间接依赖关系，具有一定的局限性。提出基于拟间接依赖关系挖掘隐变迁的方法，根据事件日志中活动间的轮廓关系构建初始模型，通过拟间接依赖关系表找出日志序列之间的约束体。利用整数线性规划方法，查找符合要求的拟间接关系变迁对，从而挖掘出拟间接关系变迁对中存在的隐变迁。将隐变迁融合到初始模型中，得到含有隐变迁的目标模型。通过具体的实例分析验证了该方法的有效性。

关键词: 隐变迁, 流程挖掘, 拟间接依赖关系, 整数线性规划, Petri网