基于0-1矩阵的频繁项集挖掘算法研究  

Research of frequent itemsets mining algorithm based on 0-1 matrix

作　　者：张月琴[1] 

机构地区：[1]南京工业大学电子与信息工程学院

出　　处：《计算机工程与设计》2009年第20期4662-4664,4667,共4页Computer Engineering and Design

摘　　要：通过对关联规则挖掘技术及经典算法Apriori的研究和分析,结合该算法存在的问题,提出了一种基于0-1矩阵的频繁集挖掘算法。该算法使扫描数据库的次数减为一次,有效率地降低I/O的存取时间,直接通过支持矩阵行向量的按位与运算来找出频繁集,而不需要Apriori算法中的自联接和减枝,提高了系统的运行效率。通过不断压缩支持矩阵,不仅减少了高次频繁集挖掘的时间,而且节约了存储空间。仿真实验表明,该算法是有效的。In view of the association rule mining technology and its classic Apriori algorithm research and analysis, combined with the existence of the algorithm, frequent itemsets mining algorithm based on 0-1 matrix is proposed. The new algorithm only needs to scan whole database once, and efficiently reduces the I/O time. Through direct support of the matrix by the vector-line operations and to identify frequent sets, it simplifies the join step and the prune step in Apriori algorithm, and improves operating efficiency. Also it reduces the support matrix unceasingly to reduce the higher mode frequent collection excavation time and save the storage space. Finally, the experimental results show the proposed method is effective.

关 键 词：数据挖掘 关联规则 频繁集挖掘 APRIORI算法 0-1矩阵 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]