基于双空间搜索的频繁项挖掘方法  

Dual Space Search Algorithms for Mining Frequents

作　　者：王晓峰[1] 王天然[2] 

机构地区：[1]沈阳化工学院计算机系,沈阳110021 [2]中国科学院沈阳自动化所,沈阳110003

出　　处：《计算机科学》2002年第4期55-60,共6页Computer Science

基　　金：辽宁省自然科学基金(9910200205);辽宁省教育厅高校科研基金(20012073)

摘　　要：1引言 1998年Roberto J.和Bayardo Jr.[7]利用自底向上搜索和项目集排序的方法建立了一种挖掘长型频繁项的Max-Miner算法;Lin D.和Z.Kedem[6]提出了一种双向钳形搜索Pincer-Search方法,利用自底向上搜索产生的非频繁项集来约束和修剪自顶向下方向的最大候选频繁项集,候选频繁项集来自于Apriori方法.这两种方法虽然细节有所不同,但修剪最大频繁项的思想类似.The famous Apriori algorithm proposed by Agrawal has become a classic algorithm for mining association rules, it is widely applied to various fields such as trade decision-making, bank evaluating credit, finance insurance etc. This approach is an effective down-top algorithm for mining frequent, but it will fall across time-consuming huge computing problems in mining long pattern frequent itemsets (as 100 items). We propose a dual space search algorithm for mining frequent items. Some of new conceptions such as transaction space and items space, mutuality set mapping and dual space search etc. are proposed, Proposed algorithms can better solve problem of mining long frequent, the validity of proposed algorithms is proved through theory analysis. Some examples of computation also given. Using transaction and itemsets information for pruning useless candidate frequent itemset, dual space search algorithm has higher efficiency than Apriori algorithm.

关 键 词：数据库 数据挖掘 数据结构 频繁项挖掘方法 双空间搜索 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]