一种新的前缀立方索引机制  

A New Index Mechanism for PrefixCube

作　　者：聂晶[1] 冯剑琳[1] 王元珍[1] 

机构地区：[1]华中科技大学计算机科学与技术学院,湖北武汉430074

出　　处：《小型微型计算机系统》2007年第3期451-455,共5页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(60303030)资助;中国博士点基金项目(20030487032)资助.

摘　　要：前缀立方在浓缩数据立方的基础上利用前缀共享和基本单元组技术有效地缩小了数据立方的尺寸.由于前缀共享产生的分组结构,浓缩数据立方的索引CuboidTree并不适合前缀立方.所以,根据前缀立方包含分组结构的特性,结合BUB-Tree和R-Tree技术对CuboidTree进行改造,最终得到了一种新的索引机制Bound-CuboidTree.实验证明,这种索引机制比已有的前缀立方索引Prefix-CuboidTree更适合前缀立方的结构.PrefixCube based on condensed cube was proposed to reduce the size of data cube more efficiently by augmenting BST condensing with prefix-sharing. Because of the group structure after prefix-sharing,the CuboidTree for Condensed cube indexing is not appropriate for PrefixCube. By introducing BUB-Tree and R-Tree technique into CuboidTree,a new index mechanism called Bound-CuboidTree was proposed. According to the results of extensive experiments,it is more efficient for PrefixCube than Prefix-CuboidTree.

关 键 词：前缀立方 分组 Bound—CuboidTree 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]