面向业务过程的概念漂移检测方法D-SPC  

Business Process Oriented Concept Drift Detection Method D-SPC

作　　者：戎小玉 赵海燕[1] 曹健[2] 陈庆奎[1] RONG Xiaoyu;ZHAO Haiyan;CAO Jian;CHEN Qingkui(Shanghai Key Lab of Modern Optical System,and Engineering Research Center of Optical Instrument and System,Ministry of Education,University of Shanghai for Science and Technology,Shanghai 200093,China;Department of Computer Science and Technology,Shanghai Jiao Tong University,Shanghai 200030,China)

机构地区：[1]上海市现代光学系统重点实验室,光学仪器与系统教育部工程研究中心,上海理工大学光电信息与计算机工程学院,上海200093 [2]上海交通大学计算机科学与技术系,上海200030

出　　处：《小型微型计算机系统》2025年第2期266-273,共8页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(62072301)资助。

摘　　要：企业内部存在着大量的业务过程,这些业务过程的模型可能会发生变化从而使得一些数据分析的算法失效.检测过程模型变化这一“概念漂移”现象对业务过程管理具有突出的意义.目前的过程模型概念漂移检测存在参数过多,特征抽取复杂等问题.本文中从过程模型的变化伴有新活动对的出现、旧活动对的消失以及活动对出现频率的变化这一基本现象出发,设计了一种简洁高效的检测框架.在该框架中,针对活动对的出现和消失设计了对活动对出现的轨迹索引进行差分的检测方法;针对活动对频率的变化,在轨迹索引差分的基础上首次运用统计过程控制的工具来检测概念漂移.为了验证本文提出方法的性能,使用合成日志和真实生活日志进行评估,并将本文方法与2种相似方法和4中不同方法进行了实验对比,实验结果表明,本文的方法有效且有更高的精度.There are a large number of business processes within an enterprise,and the models of these processes may change,rendering some data analysis algorithms ineffective.Detecting the"conceptual drift"phenomenon of process model changes is of great importance for business process management.The current process model concept drift detection has too many parameters and complex feature extraction problems.In this paper,we design a simple and efficient detection framework based on the basic phenomenon that process model changes are accompanied by the appearance of new activity pairs,the disappearance of old activity pairs,and changes in the frequency of activity pairs.In this framework,a detection method is designed to differ the trajectory indexes of the emergence and disappearance of activity pairs,and a statistical process control tool is applied for the first time to detect concept drift based on the trajectory index differencing for the change of activity pair frequency.In order to verify the performance of the proposed method in this article,synthetic logs and real life logs were used for evaluation,and experimental comparisons were made with two similar methods and four different methods.The experimental results showed that the proposed method is effective and has higher accuracy.

关 键 词：过程挖掘 概念漂移 差分分析 统计过程控制 变更点检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]