摘要
对分布式数据库多层关联规则挖掘的理论和方法进行了研究,提出了一种基于频繁模式树FP-tree(FreguentPatterntree)的快速挖掘算法DMAML_FPT(DistributedMiningAlgorithmofMultipleLevelbasedonFP-tree)。与类Apriori算法相比较,该算法最多只需扫描数据库三遍,不需产生和传输大量的候选项集,减少了数据通信量,从而提高了数据挖掘的效率。实验结果表明算法DMAML_FPT是可行和有效的。
A fast mining algorithm named DMAML_FPT( Distributed Mining Algorithm of Multiple Level based on Frequent Pattern Tree) was presented after researching multi-level association rules mining in distributed database. Comparing with Apriori-like algorithms, DMAML_FPT only need to scan the database for three times, eliminated the need for generating the candidate items, reduced the communication cost, and improved efficiency of data mining. Experiment results show that the algorithm is feasible and efficient.
出处
《计算机应用》
CSCD
北大核心
2005年第12期2858-2861,共4页
journal of Computer Applications
基金
国家自然科学资助基金(70371015)
关键词
数据挖掘
分布式数据库
多层关联规则
频繁模式树
data mining
distributed database
multi-level association rules
frequent pattern tree(FP-tree)