基于数组的关联规则算法的改进  

New Optimization Association Rule Algorithm Based on Array

作　　者：李瑞[1] 康良玉[1] 耿浩[1] 

机构地区：[1]大连交通大学软件学院,大连116028

出　　处：《科学技术与工程》2008年第21期5846-5849,5857,共5页Science Technology and Engineering

摘　　要：针对Apriori算法的不足,提出了基于数组的关联规则挖掘算法,该算法只扫描数据库一次,将数据库中的数据存于数组向量中。基于关联规则的性质,对扫描的事务数和项目数,连接步骤等进行压缩和优化,并且利用一维数组对候选2-项集进行计数,从而避免大量候选2项集的产生,有效解决了传统算法候选2项集的瓶颈问题,此算法与Apriori算法相比有明显的提高。For the insufficient of Apriori, a new association rule malgorithm based on arraies is proposed. It scans the database only once, and puts data in the database into different array vectors, based on the properties of Association Rule , this algorithm compresses the scanned transactions and items, improves join process, it also uses a one dimensional array to count candidate 2-items in the database , and it avoids mass generation of candidat 2- itemset , solves the traditional bottlenecks problems about candidate2-itemset effectively o compared with traditional Apriori algorithm, this algorithm has obvious improvement in efficiency.

关 键 词：关联规则 频繁项集 数据挖掘 数据库 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]