快速挖掘加权频繁项集的矩阵位串算法  

Efficient algorithm of mining weighted frequent itemsets based on matrix and bit string

作　　者：李娟[1] 张明义[1] 汪维清[1] 

机构地区：[1]西南大学计算机与信息科学学院,重庆400715

出　　处：《计算机工程与设计》2007年第11期2533-2536,共4页Computer Engineering and Design

摘　　要：关联规则挖掘的应用日益广泛,但已经提出的大多关联规则挖掘算法都是把数据仓库中各个项目按平等一致的方式加以处理的。然而,在现实世界中,不同的项目往往有着不同的重要性。现有的有关加权关联规则的研究中,大多采用的加权方法不太好,或挖掘算法效率不够高。为此,提出了一种新的挖掘加权关联规则的算法,该算法采用矩阵和位串技术,只需要对数据库扫描一遍,可快速挖掘出所有的加权频繁项集,并且存放辅助信息所需要的空间也较少。研究表明该算法比已有的算法更高效。Association rules are useful in many areas. Most of the presented algorithms treat each item as uniformity. However, in real world databases, different items usually have different importance. And in the current researches of weighted association rules, most algorithms don＇t have good method to set weight value or perfect performance. So, a novel algorithm based on matrix and bit string for finding weighted frequent itemsets is presented. This algorithm only needs scan the database one time, and occupies little memory for assistant information. Researches indicate that the new algorithm is more efficiency compared with the presented ones.

关 键 词：数据挖掘 关联规则 加权关联规则 频繁项集 最小支持期望 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]