基于定制时间约束的正负时态关联规则挖掘

传统关联规则挖掘是在整个事务数据库的时间范围内进行的,但有时用户想得到某一特定时间范围(如商品的促销阶段)内的关联规则,该文对这一问题进行了详细讨论,提出了基于定制时间的时态支持度、时态频繁项集、时态置信度、时态关联规则等概念,在传统Apriori算法的基础上提出了挖掘时态频繁项集的算法。另一方面,讨论了当同时考虑正、负关联规则出现的矛盾规则问题以及用相关性解决这一问题的方法,提出了挖掘正负时态关联规则的算法,实例说明了算法的执行过程及有效性。

Customized Interval based Method for Mining Positive and Negative Temporal Association Rules

The association rules are discovered traditionally in the interval of the whole transaction database.The users,however,sometimes are interested in the rules in a special interval(the sales promotion period for example).This prob-lem is discussed in details.Some concepts,such as temporal support,temporal frequent itemsets,temporal confidence and temporal association rule,are proposed based on customized interval.The traditional algorithm Apriori is modified to dis-cover temporal frequent itemsets.On the other hand,self-conflicting rules maybe occur when studying positive and nega-tive association rules simultaneously.This problem is discussed and the corresponding solution is given by applying cor-relation in association rules.An algorithm is proposed to mine positive and negative temporal association rules.An exam-ple is also given to demonstrate the algorithms' efficiency.