Abstract: A large amount of lowclass granularity data is existed in data warehouse, and the storage strategies of them are closely related to the performance and efficiency of on line analytical processing and data mining. After analyzing the types and characteristics of lowclass granularity dada in data warehouse, this paper discusses various static data division strategies, and studies some dynamic data division strategies caused by the variety from structure to content in detail, then a dynamic data division technique based on attributes compatibility and semantic equivalency is proposed.

Key words: Data warehouse, Granularity partition, Data division, Attributes compatibility, Semantic equivalency

摘要： 数据仓库中存在着巨大容量的低粒度数据，其存储策略的好坏直接影响到联机分析处理和数据挖掘的性能及效率。该文分析了数据仓库系统中数据分割的一般原则，详细论述了静态数据分割的各种策略，并对结构和内容两个方面的变化所引起的动态数据分割策略进行了详细研究，提出了基于属性相容和属性语义等价的动态数据分割技术。

关键词: 数据仓库, 粒度划分, 数据分割, 属性相容, 语义等价

CLC Number: 

• TP311