基于对齐的BPMN 2.0模型符合性检测算法

发布时间：2018-05-30 02:20

  本文选题：BPMN + .模型　； 参考：《计算机研究与发展》2017年09期

【摘要】：符合性检测方法作为比较和关联事件日志与流程模型的技术,是三大核心流程挖掘技术之一,可用于量化符合性和诊断偏差.BPMN 2.0模型具有丰富的表达能力,能够表达多实例、子流程、边界事件、OR网关等多种复杂模式,但是目前还没有针对这些复杂模式的BPMN 2.0模型符合性检测算法.针对该问题,提出了基于对齐的BPMN 2.0模型符合性检测算法Acorn,该算法支持上述多种复杂模式.在深入分析BPMN 2.0模型中多种复杂模式的具体语义并分析其具体使能情况的基础上,Acorn算法引入对齐操作,利用A*搜索算法寻找到代价最小的匹配轨迹,同时引入虚拟代价和预估代价来对A*算法进行搜索空间的优化,最后根据最佳匹配轨迹来计算模型与日志的契合度.实验表明,Acorn算法能够正确有效地计算带有复杂模式的BPMN 2.0模型与日志之间的契合度,且虚拟代价和预估代价的引入,大大减少了搜索空间,有效提高了算法的运行速度.
[Abstract]:As a technique of comparing and associating event log and process model, conformance detection method is one of the three core process mining techniques. It can be used to quantify compliance and diagnose deviation. BPMN 2.0 model has rich expression ability and can express many examples. There are many complex patterns such as sub-flow, boundary event OR gateway and so on, but there is no BPMN 2.0 model conformance detection algorithm for these complex patterns. In order to solve this problem, an alignment based BPMN 2.0 model conformance detection algorithm (Acorn) is proposed, which supports many complex patterns mentioned above. On the basis of deep analysis of the semantics of many complex patterns in the BPMN 2.0 model and the analysis of its specific enabling situation, this paper introduces alignment operations into the Acorn algorithm, and uses the A * search algorithm to find the least costly matching locus. At the same time, the virtual cost and the estimated cost are introduced to optimize the search space of the A * algorithm. Finally, the consistency between the model and the log is calculated according to the optimal matching trajectory. The experimental results show that the algorithm can correctly and effectively calculate the consistency between the BPMN 2.0 model with complex schema and the log, and the introduction of virtual cost and prediction cost greatly reduces the search space and effectively improves the running speed of the algorithm.
【作者单位】： 清华大学软件学院;首都经济贸易大学信息学院;
【基金】：国家重点研发计划项目(2016YFB1001101) 国家自然科学基金项目(61472207,61325008,61402301)~~
【分类号】：TP301.6
，

本文编号：1953485