关联规则挖掘中的规则等价与精简

Rules Equivalent and Streamlining in Association Rules Mining

作为推荐系统的一种模式,购物篮分析不同于基于用户的和基于项目的协同过滤推荐,而是根据商品在购物篮中的组合进行聚类分析,使用关联规则挖掘算法得到关联规则集并用于推荐商品。利用提升度公式可推导出两种关联规则的等价关系,对Apriori算法得到的关联规则集作冗余精简处理,实验表明精简率可达到60%以上。

Abstract： As a model of the recommender systems, market basket analysis is different from two recommender systems which are both user-based and item-based collaboration filtering. Its cluster analysis is based on the combination of goods in the shopping basket, and the association rules set are obtained by the association rule mining algorithm to recommend products. The formula of Lift can be used to deduce the two equivalence relationships of the association rules, and the redundancy association rule sets which are obtained through the Apriori algorithm can be streamlined. The relevant experiments show that streamlining rate can reach over 60%.