摘要
数据立方体计算是联机分析中一项重要的技术。研完工作者提出了多种具有不同存储空间和查询相应时间的数据立方体,每种数据立方体有独自的生成算法。本文分析了使用关系系统作为存储结构的一般数据立方体、部分数据立方体和浓缩数据立方体的原理,提出用合作伙伴的概念统一这三类数据立方体的思想,并设计了一个算法TCUBE用于生成这些数据立方体。我们使用了一个实际数据集测试了TCUBE的性能,结果表明它生成浓缩数据立方体的速度要快于原有的算法。
The computation of data cube is one important technique in On-Line Analytical Processing. Researchers have proposed many kinds of data cubes that are of different query response time and occupy varying size of space. Any more, each data cube has its own constructing algorithm. This paper analyzes the principles of normal data cube, partial data cube and condensed data cube that put their tuples into relation system, proposes to use the idea of fellowship to unify these kinds of data cube, and designs an algorithm TCUBE to obtain them. We also conduct an experiment using a real data set to verify the performance of TCUBE. The results show that TCUBE outperforms the original algorithms used to produce condensed cube.
出处
《计算机科学》
CSCD
北大核心
2004年第10期93-95,共3页
Computer Science