Abstract: To improve the low efficiency of the existing methods when judging whether traces are compliant or not,a method to achieve business process conformance checking was proposed based on an activity-order decision trees,and then the tree was pruned to improve detection efficiency.The fit traces were visited in the event log,and the attribute values of activities were recorded according to the order of activities,then a decision tree was constructed taking the activity order as the attribute.With the guarantee of the precision,the decision tree was pruned to reduce its structural scale and improve the judgement efficiency,and the fitness of the traces was judged in the test set.Simulation experiments were implemented with the earthquake emergency plan process and the real life cases,which verified that the algorithm significantly had improved the business process conformance checking efficiency compared with the string distance metric based method and the classical alignment method.Based on the experimental results of real event cases,the feasibility and superiority of the proposed method in business process conformance checking were further illustrated.

Key words: conformance checking, earthquake emergency processes, decision tree, event logs, pruning

摘要： 针对现有方法判断轨迹合规情况效率较低的问题,提出了一种基于活动次序决策树检查业务过程合规性的方法,并对该树进行剪枝处理以提高检测效率。首先,访问事件日志中完全拟合的轨迹,按照活动次序分别记录下每个活动的可取值,构建一棵以活动次序为属性的决策树;然后,为了减小决策树的结构规模提高判断效率,在尽量保证准确率的同时,对决策树进行剪枝,并对测试集中轨迹的拟合情况进行判断;最后,分别以地震应急流程及真实事件案例为数据集进行仿真实验。相较基于字符串距离度量方法和经典对齐方法,该算法显著提高了业务过程合规性检查效率,基于真实事件案例实验结果,进一步说明了该方法在进行业务过程合规性检查时的可行性和优越性。

关键词: 合规性检查, 地震应急流程, 决策树, 事件日志, 剪枝