基于位对象的最大频繁模式挖掘算法  

Algorithm Based on Bit Objects for Mining Maximal Frequent Patterns

作　　者：杨君锐[1] 张铁军[1] 刘南艳[1] 

机构地区：[1]西安科技大学计算机科学与技术系,陕西西安710054

出　　处：《西南交通大学学报》2008年第4期488-493,共6页Journal of Southwest Jiaotong University

基　　金：陕西省自然科学基金资助项目(2005F13);陕西省教育厅专项科研基金资助项目(06JK248)

摘　　要：提出了基于位对象的最大频繁模式挖掘算法.算法中,用位对象表示数据,并用位对象概念改进FP-Tree.用深度优先搜索策略,通过压缩数据库,并用位对象的特性简化模式支持度的计数,使挖掘时不需产生条件FP-Tree和候选项目集,以提高最大频繁模式的挖掘效率.实验结果验证了BFP-Miner的有效性.A new algorithm based on bit objects, BFP-Miner, for mining maximal frequent patterns was proposed. It uses the bit objects to express data and to improve the FP-Tree （frequent pattern tree）. The algorithm uses depth-first search strategy, and simplifies the support counting of frequent patterns with the characteristics of the bit objects and by compression of the database. Neither a conditional FP-Tree nor candidate patterns are generated during mining the maximal frequent patterns, so that the mining efficiency is increased. Experimental result verifies the efficiency of the BFP-Miner.

关 键 词：数据挖掘 关联规则 最大频繁模式 位对象 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]