基于关系矩阵的关联规则增量式更新

关联规则是当前数据挖掘研究的主要模式之一.本文提出了一种高效的增量式关联规则的挖掘算法USLIG,以处理当最小支持度改变时相应的关联规则的更新问题.该算法通过构建向量之间的关系矩阵,将频繁项目集的产生过程转化为项目集的关系矩阵中向量的运算过程,能充分利用以前的挖掘结果,只需扫描比数据库小得多的向量,克服了IUA及相关算法需多次扫描数据库的缺点.

Fast Algorithm for Updating Association Rules based on Relationship Matrix

Mining association rules is an important task for knowledge discovery. In this paper, an efficient algorithm USLIG is proposed in order to deal with the rules updating as the minimum support threshold changed. The incremental updating technique constructs the relationship matrix on vectors to indicate the association between items, and then generates frequent itemsets hereby. The algorithm can maintain the discovered association rules, which simply scans the vectors extremely smaller than database and outperforms IUA and other algorithms that need to make multiple passes over the large database.