Abstract: Aiming at the business processes to be optimized,a solution based on association rules mining of database logs was proposed.In this solution,the database logs were turned into geometric-measurable workflow logs through vectorization,and the similarity between database logs was comprehensively analyzed in angle and distance.In this way,the user behavior profiles were obtained with similarity-based correlation analysis algorithm,which could guide the nodes splitting and merging as well as reshape the node structure to optimize the processes.Meanwhile,this solution also assessed the accuracy of correlation analysis by multi-order iteration and made the algorithm be executed in a reasonable range.

Key words: similarity analysis, association rules mining, process optimization

摘要： 针对需要优化的业务流程,提出基于数据库日志之间关联规则挖掘的解决方法。通过对数据库日志向量化使其变为可几何度量的流程日志,并从角度和距离两方面综合分析数据库日志的相似性。采用基于相似性的关联分析算法获得用户行为模式以指导节点的分裂或合并,实现节点结构重塑从而优化流程。该方法还通过多阶迭代的方式评价关联分析的准确性,使算法可以在合理范围内执行。

关键词: 相似性分析, 关联规则挖掘, 流程优化