一种高精确度的过程模型修复方法

High-precision repairing method of process models

为了解决现有方法修复的过程模型精确度不高的问题,提出一种高精确度过程模型修复方法。为了便于确定偏差的位置,基于Petri网可达标识提出扩展校准的概念。针对扩展校准中的日志动作收集形成子日志,并将子日志挖掘出的子过程插入原模型中,避免了现有方法由于添加自环导致这一子过程多次重复发生的问题。结合Petri网的过程树,通过查找过程树的非叶子节点,能够定位到Petri网的选择结构。针对选择结构提出一种新的偏差类型,并给出判定方法,将挖掘出的子过程作为选择结构的一条分支,对模型进行修复。通过青岛某医院就诊数据的实例分析,验证了该修复算法有较高的拟合度和精确度。

To solve the problem of low precision for repairing process model with existing method, a high precision repairing method was proposed. The extended alignments were proposed based on reachable marking of Petri nets to confirm location of deviations. For log moves in extended alignments, the sublogs were collected, and the subprocesses mined by sublogs were inserted into an original model. The method avoided that the subprocesses were added to the original model at the right location by loops. Combined with the process trees of Petri nets, the choice structure of Petri nets could be identified easily by searching non-leaf nodes. A new type of deviation was proposed for choice structures. The method of judging deviations and collecting corresponding sublogs was proposed. Mined subprocesses were added to a proper choice structure of model as some branches. The fitness and precision the proposed methods were illustrated by an example of medical processes data in a hospital of Qingdao.