基于频繁活动序列挖掘的过程改进机会分析

Identification of Process Improvement Opportunity Based on Frequent Activity Sequential Pattern Mining Technology

为了指导质量管理过程的持续改进,提出了基于图结构的工作流频繁活动序列模式的挖掘方法.采用基于Apriori方法的频繁活动序列挖掘算法,认为k-频繁图集中,当一个图减去其中的一个源顶点后,如果所得到的图与另一个图减去其中的一个沉顶点后的图相同时,可以连接生成一个(k+1)-候选频繁图,从而减少了传统Apriori算法迭代过程中生成的冗余候选频繁图的数目.文中以某飞机制造公司的质量外审意见处理流程为例,对改进Apriori算法的应用效果进行了验证,结果表明,该方法能够有效地挖掘出历史过程实例集中所蕴含的频繁活动执行序列,辅助企业可从过程组成的角度来寻找质量管理过程的改进机会.

To guide the continuous improvement of quality management process by analyzing history workflow instance logs, a method for identifying frequent activity sequential pattern related with given quality target is presented. Aiming at the multiple loop structures in quality management process network,an improved graph pattern mining approach based on Apriori algorithm is put forward, where a frequent pattern of size k, denoted as Gi, is joined with another frequent pattern of size k, denoted as Gj, to form a candidate frequent pattern of size （k ＋1）,if there exists a source vertex s in Gi and a sink vertex e in Gj such that the graph derived from Gi by subtracting node s equals with the graph derived from Gj by subtracting node e. This coalition operator only considers the source vertexes and sink vertexes, which reduces the amount of redundant frequent candidates generated in the traditional Apriori iterations. The external audit treatment process is used as a case study to illustrate the operation of the proposed method, and the result shows the effectiveness to discover the frequent activity sequential patterns to lead to the improvement of quality management process.