过程挖掘中增强活动依赖的最优对齐算法

OPTIMAL ALIGNMENT CALCULATION METHOD TOENHANCE ACTIVITY DEPENDENCY IN PROCESS MINING

一致性检验是过程挖掘领域中检验日志与模型之间偏差的有效方法,对齐是众多先进方法之一。现阶段最优对齐的成本计算大多只与对齐中移动个数保持相关,缺乏对单一活动成本的考虑。因此,提出一种增强活动依赖的最优对齐加权计算方法。基于动态规划求解日志与模型之间的序列对齐;统计合法移动信息,差异化活动及移动类别的成本进行依赖增强;依据对齐成本加权计算方法计算最优对齐并以Java应用的形式对以上方法进行实现。实验利用一个常规化模型及其部分日志对方法的可行性及有效性进行了验证。

Conformance checking is an effective method for examining deviations between logs and models in the field of process mining,and alignment is one of many advanced methods.Most of the cost calculations of optimal alignment at this stage only keep correlation with the number of moves in the alignment and lack consideration of the cost of a single activity.We proposed an optimal alignment weighted calculation method to enhance the activity dependence.We solved the sequence alignment between the log and the model based on dynamic planning.We statistically calculated the cost of legal moves,differentiated activities and moved categories to enhance the dependency.We calculated the optimal alignment based on the alignment cost weighted method,and implemented the above method as a Java application.The experiments used a generalized model and some of its logs to prove the feasibility and validity of the method.