一个改进的并行关联规则算法研究  

An Improved Parallel Algorithm for Mining Association Rules

作　　者：陈涛[1] 张玮[1] 

机构地区：[1]安徽财经大学信息工程学院,安徽蚌埠233041

出　　处：《计算机技术与发展》2007年第1期139-141,共3页Computer Technology and Development

基　　金：安徽省教育厅自然科学科研基金(2005KJ051);安徽财经大学教研课题(ACJYZD200510);安徽财经大学信息工程学院青年基金

摘　　要：在研究关联规则挖掘算法的基础上,对并行关联规则算法进行了比较全面的分析,并给出了并行数据挖掘的计算框架。提出了一个以计算服务器为中心节点的并行挖掘算法,可以发挥各局部节点的优势,无需各局部节点进行通信,减少了各局部节点的通信负荷。通过理论分析和实验数据验证,该算法具有较好的可扩展性和海量处理能力,特别是在节点数目较多的情况下更显示出优势。Based on investigation of mining association algorithms rules, the paper analyzes parallel association rules algorithm in details. At the same time, the parallel calculating architecture of data mining is given. To overecome problems existing in performance and efficiency, algorithm for parallel mining association rules based on computing server as the central processor is put forward,which can develop each advantage of local processor and reduce the cost of communication among loal processor. Theoretic analysis and experimental result show that it holds better calibility and ability to deal with massive data, especially under the condition of the more local processor number.

关 键 词：数据挖掘 关联规则 并行处理 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]