一种基于BUC的水平加权关联规则挖掘算法  

A BUC-BASED MINING ALGORITHM FOR HORIZONTAL WEIGHTED ASSOCIATION RULES

作　　者：王斌[1] 丁祥武[1] 

机构地区：[1]东华大学计算机科学与技术学院,上海201620

出　　处：《计算机应用与软件》2008年第12期112-115,共4页Computer Applications and Software

基　　金：上海市科委项目资助(05DZ11C06)

摘　　要：关联规则挖掘可以从大量数据中发现项集间潜在而有趣的相互联系。针对用户对每个项目感兴趣的程度不同,一些学者提出了水平加权关联规则。然而每次生成新候选集后对整个数据库事物的扫描成为算法效率的一大瓶颈。为进一步提高加权关联规则的挖掘效率,在原有的水平加权关联规则算法的基础上,采取了深度优先的策略,提出了一种基于BUC的水平加权关联规则挖掘算法——BUC-MINWAL。改进算法可以大大减少对数据库的扫描范围。实验结果表明,改进的算法有更好的执行效率。Association rules mining can find out latent and interesting rules form large volumes of data. According to the fact that customers pay different attention to different items, some scholars proposed algorithms for association rules with horizontal weighted items. But every time after forming a new candidate items group, the scan of whole transactions in database is still the bottleneck of the algorithm efficiency. In order to enhance the performance of mining, in this article we imbue the previous algorithms with depth-first strategy and proposes a mining algorithm for association rules with horizontal weighted items on bottom-up computing （BUC） basis, which is called BUC-MINWAL. This im- proved algorithm can greatly reduce the scanning scope of database. The results of experiments show that the improved algorithm has better performance in efficiency.

关 键 词：数据挖掘 加权关联规则 自底向上运算 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论] O661.1[自动化与计算机技术—计算机科学与技术]