一种基于线性链表的关联规则挖掘算法  

An Association Rules Mining Algorithm Based on Linear Linker

作　　者：李晓虹[1] 杨有[1] 

机构地区：[1]重庆师范大学数学与计算科学学院,重庆400047

出　　处：《计算机科学》2007年第9期142-144,共3页Computer Science

基　　金：重庆师范大学科研项目资助(编号:06XLY026)

摘　　要：关联规则挖掘是数据挖掘的一个重要研究方向,其算法主要有Apriori算法和FP-growth算法,它们需要多次扫描事务数据库,严重影响算法的效率。为了减少扫描事务数据库的次数,本文提出一种基于线性链表(Linear Linker)的LL算法,它只需扫描事务数据库一次,把事务数据库转换为线性链表LL,进而对LL进行关联规则挖掘。实验表明,LL算法的时间开销明显优于Apriori算法和FP-growth算法,且LL算法通过定义备用候选频繁项目集,有效地支持了关联规则的更新挖掘。Association rules mining is an important research aspect of data mining. Apriori algorithm and FP-growth algorithm are the main mining algorithms of this aspect. Because of multiple scanning of transaction database, these algorithms＇ performances are restricted heavily. An algorithm which was based on linear linker, we called LL algorithm, was presented for the reducing of scanning. LL algorithm scanned the transaction database only once. It translated the transaction database to a linear linker and mined the linear linker further more. The Experiments showed that LL algorithm＇s time cost is less than Apriori＇s and FP-growth＇s evidently. Furthermore, LL algorithm could support incremental updating effectively by employing the concept of backup candidate frequent itemset.

关 键 词：关联规则挖掘 更新挖掘 线性链表 事务数据库 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]