一种支持在线数据挖掘的存储结构

A Stored Structure to Support Online Data Mining

数据挖掘是一个庞大的计算过程,频繁项集挖掘是数据挖掘中很重要的一部分。本文提出一种基于数据挖掘的存储频繁项集结构——BFp-tree,对频繁项集进行预处理,并将其存储在磁盘上,以支持在线挖掘要求。BFp-tree利用共享前缀和后缀的特点节省存储空间,对稠密的数据集压缩效率较高。

Frequent itemset mining is an important part of data mining ,which is a computation intensive process. This paper propose a data mining based structure-BFp-tree to store frequent itemsets. This structure store precomputed frequent itemsets on a disk to support online mining requests. To save space, share the prefix and suffix in BFp-tree structure is applied. BFp-tree is efficient in dense dataset compressing.