基于流程案例簇的任务关系挖掘算法

Mining algorithm of task dependencies based on process case clusters

为了进行流程发现,提出一种基于流程案例簇的任务关系挖掘方法。该方法首先将基本案例按照特征向量分为多个案例簇,根据基本案例的任务轨迹对案例簇中任务间的依赖关系进行挖掘;然后给出了基于循环基元的循环结构建模和挖掘方法,最终可以从事件日志导出流程中完整的任务依赖关系以及存在的循环结构。所提方法能正确处理任务依赖关系随案例属性取值不同而变化的情况。只要日志完备,基于挖掘到的任务关系和循环结构就能得到一个与原流程行为等价的流程模型。

To make process discovery, a case cluster-based derivation method of task dependencies was presented. The basic cases were classified into case clusters according to feature vectors. According to task traces of basic ca- ses, task dependencies among case clusters were mined, and a modeling and discovery approach of loop structures based on loop-unit was proposed. All the task dependencies and loop structures in processes were generated from e- vent logs. Compared to the existing methods, the proposed method could correctly handle such situations that task dependencies change with case attribute values. Based on the mined task dependencies and loop structure, a behav- ior-equivalent process model was obtained as long as the event log was complete.