漏洞维及其相关操作的研究  

A Simple and Economical Remedy for Existing OLAP Systems That Can Not Deal Effectively with Ragged Dimensions

作　　者：袁霖[1] 李战怀[1] 

机构地区：[1]西北工业大学计算机学院,陕西西安710072

出　　处：《西北工业大学学报》2004年第1期1-5,共5页Journal of Northwestern Polytechnical University

基　　金：国家自然科学基金 (6 0 0 730 5 5 );西北工业大学博士论文创新基金资助

摘　　要：维是面向 OLAP应用的多维数据模型的必要组成部分。维模式是等级的集合 ,不同的等级使用户可以对不同粒度的度量数据进行观察。文中的研究对象是一类特殊结构的维 ,其中某些子成员的父成员不在其直接的上一个等级之中 ,相当于维层次结构中出现漏洞 ,当度量聚集到带有漏洞的等级时 ,很可能导致错误的计算结果 ,从而引发错误的决策。通过在传统的多维数据模型之上增加一个查询处理层 ,以保证多维操作语义的正确性及对查询优化的支持。现有的Microsoft′s OLAP (On Line Analytical Processing) system can deal effectively with ragged dimensions. But many existing OLAP systems can not; if they must be reconstructed to remove this defect, the cost is high. We propose a simple and economical remedy to enable existing OLAP systems to deal effectively with ragged dimensions. This paper focuses on ragged dimensions in multi dimensional data model (MDDM) for OLAP. Logical parents of some members of a ragged dimension are located two or more levels above these members. In other words, there exist empty holes in dimension hierarchy and, as a result, aggregation of measure data computed directly from lower levels may be incorrect. Section 3 explains in some detail how to extend the traditional MDDM by representing a ragged dimension as a special kind of conventional dimension. On the basis of this special kind of conventional dimension, section 4 explains how to add a query processing level to existing OLAP system. This query processing level is a simple and economical remedy which can enable an existing OLAP system to deal effectively with ragged dimensions.

关 键 词：漏洞维 多维数据模型 联机分析处理 OLAP 数据仓库 维模式 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]