一种基于序列数的关联规则挖掘算法  

An Algorithm of Association Rule Mining Based on Sequence Number

作　　者：王仕平[1] 蒋玲[2] 熊江[3] 方刚[3] 

机构地区：[1]成都大学信息科学与技术学院,成都610106 [2]成都大学实验技术中心,成都610106 [3]重庆三峡学院计算机科学系,重庆404000

出　　处：《西南大学学报（自然科学版）》2011年第3期122-127,共6页Journal of Southwest University(Natural Science Edition)

基　　金：重庆市教委科技资助项目(KJ091108)

摘　　要：在关联规则挖掘研究中,为了在产生候选频繁项时减少算法存在的重复计算和冗余候选项,为了在计算支持数时减少扫描事务数据库的次数,提出了一种基于序列数的关联规则挖掘算法,其关联规则适合挖掘任何长度.该算法用事务属性的布尔约简法,将传统事务数据转换成二进制数,然后用数字的递增和递减两种方式双向搜索候选频繁项;算法通过序列数的度来计算支持数,实现一次扫描数据库,有效地提高了算法的效率.In association rule mining research,in order to reduce repeated computing and redundant candidate of algorithm when generating candidate frequent itemsets,and to reduce the time of repeated scanning of transaction database when computing support,an algorithm of association rule mining based on sequence number is presented,which is suitable for mining any long association rules.The algorithm uses Boolean reduction of transaction attributes to turn traditional transaction data into binary,and then uses two ways of number ascending and descending to double the search candidate frequent itemsets.The algorithm computes support by sequence number degree to realize database one scanning,thus enhancing its efficiency.The experimental data indicates that this algorithm is faster and more efficient than similar algorithms.

关 键 词：数据挖掘 关联规则 布尔约简法 双向搜索 序列数 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]