关键词: Galois联络 关联规则 数据挖掘 MNRM算法

Abstract:

Since the association rule＇s mining is NP-hard,the key is how to reduce frequent itemsets.In the paper,the generating basis for exact association rul  es and the proper basis for approximate association rules are addressed based on the Galois connection,and the Galois closure properties.This paper pres ents a new algorithm called MNRM to discover minimal non-redundant exact and approximate rules.Compared with the Apriori algorithm,these rules are the m ost non-redundant,the computing complexity is reduced.And these rules have many strongpoints,such as minimal antecedents and maximal consequents,the mos t relevant association rules,limiting the number of rules produced without information loss,the most user-useful and user-relevant.

Key words: （Galois connection, association rules, data mining, MNRM algorithm）