基于数据库划分的关联规则算法  

Association rules arithmetic based on database partition

作　　者：黄学平[1] 薛安荣[1] 

机构地区：[1]江苏大学计算机科学与通信工程学院,江苏镇江212013

出　　处：《计算机工程与设计》2008年第12期3005-3007,3015,共4页Computer Engineering and Design

基　　金：国家自然科学基金项目(60603041);江苏省高校自然科学基金项目(D5KJB520017)

摘　　要：关联规则是数据挖掘的一个重要研究方向。分析了FP算法的优缺点,提出了一种基于数据库划分的算法PFp算法,从理论上证明了该算法的正确性。该方法将事务数据库划分为子事务数据库,在子事务数据库中挖掘局部频繁项集,并入到全局频繁项集中,采用连接和剪枝策略有效挖掘出局部不频繁但全局频繁的频繁项集。实验结果表明,该算法比FP算法更加有效。Association rules are an important research aspect of data mining. The traditional FP arithrnetic＇s advantage and disadvantage are analyzed, and a arithmetic called PFp based on partition is presented. It was proved to be right in the theory. This method divided the database into several sub databases, and mine the local frequent item sets in the local database, then insert the local frequent item sets into the global frequent item sets. Pruning strategy is used to mine the global frequent item sets which are not frequent in the local database. The experience shows that the PFp is more effective than FP arithmetic.

关 键 词：关联规则 频繁项集 FP树 子事务数据库 剪枝 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]