一种基于矩阵的多值关联规则的挖掘算法  

A Quantitative Association Rules Mining Algorithm Based on Matrix

作　　者：李国雁[1] 沈夏炯[1] 

机构地区：[1]河南大学计算机与信息工程学院

出　　处：《计算机工程与科学》2008年第5期72-74,77,共4页Computer Engineering & Science

摘　　要：关联规则是数据挖掘研究的主要模式之一,其中布尔型关联规则的挖掘已经有比较成熟的系统和方法,而多值关联规则的挖掘则不然。本文提出的QARMM算法利用矩阵存储数据,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,同时克服了SLIG算法和矩阵算法不能挖掘多值关联规则的弱点,只需运行一次便可挖掘出所有关联规则。实验证明,在等价的数据集上挖掘关联规则,QARMM算法比Apriori算法具有更高的效率。Association rule is one of the most important patterns for the research of data mining, and there are now mature systems and approaches for mining the Boolean association rules while there are the opposite conditions for mining quantitative association rules. This paper presents a new algorithm QARMM which makes use of matrix to store data and transforms the process of generating frequent item sets to the process of calculating matrix vectors. The QARMM algorithm overcomes the drawbacks of the SL IG algorithm, needs to run only once to mine all the association rules,and has better efficiency than the Apriori algorithm on equivalent datasets.

关 键 词：多值关联规则 可辨识向量 频繁项集集合 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]