Abstract: In this paper,the scope of discovery of the association rules is extended from single level to multiple level based on vertical data layout, and a top down and progressively deepening algorithm is put forward for discovering association rules. This algorithm calculates the frequent itemsets using equivalence classes and has no need of Hash data structure. As a result. the algorithm decreases the cost of pattern matching and speeds up the mining procedures.

摘要： 将基于垂直数据分布的关联规则的发现从单层概念扩展到多层概念，提出了自顶向下的、用等价类生成频繁项目集的发现算法，无需复杂的Ｈａｓｈ数据结构。该算法减少了项目的匹配计算，提高了挖掘的效率。

关键词: 数据挖掘, 关联规则, 等价类, 概念层次

CLC Number: 

• TP311