【作者】 张雅茜；

【导师】 夏秀峰； 李常山；

【作者基本信息】 沈阳航空工业学院 ， 计算机应用技术， 2006， 硕士

【摘要】 独立型数据集市以投入资金少及设计周期短等特点成为大多数企业可接受的快速实现企业辅助决策分析的解决方案。本文围绕独立型数据集市的体系结构、维度建模、数据转换及OLAP实现等关键技术进行了深入细致地研究，提出了扩展独立型数据集市的体系结构，并以此为基础，提出了扩展独立型数据集市中一致性维度创建、维度模型的粒度划分及数据转换等策略。 针对传统独立型数据集市之间数据不一致且难以集成为企业级数据仓库的问题，本文提出了DB-EDMS两层体系结构。在介绍维度建模的基本概念和维度模型的结构的基础上，提出了在扩展数据集市的数据集市控制中心创建一致性维度的三种策略及基于时间轮转模型进行维度模型粒度划分的基本方法。在此基础上，介绍了“沈阳市特种设备监督检验决策支持系统”维度模型的创建方法和过程。数据转换是用户能否得到高质量信息的前提和保证，本文详细讨论了数据集市中数据转换的过程和方法，提出了一种基于维度模型的数据转换模型。在研究了基于维度模型的独立型数据集市多阶段数据转换策略和管道数据转换策略的基础上，提出了以XML技术实现异构环境下独立型数据集市的多阶段数据转换算法。 最后，本文深入地研究了OLAP的产生背景、与OLTP之间关系及基本分析方法等内容，并以“沈阳市特种设备监督检验决策支持系统”检验收费多维模型为例展示了OLAP的设计和实现过程。更多还原

【Abstract】 With less investment and short development cycle, independent data marts have become the acceptable and fast solution scheme of implementing decision support system for most enterprises. After studying some key techniques such as architecture design of independent data marts, dimensional modeling, data transformation, the realization of OLAP, the architecture of extended independent data marts is proposed in this dissertation. Then strategies of establishing consistency dimensionality, dividing up granularity of dimensional model and data transformation of extended independent data marts are proposed.Focusing on the problem of data inconsistency among several traditional independent data marts and being hard to integrate into enterprise data warehouse, this dissertation proposes the two-levels architecture. After introducing the basic conception of dimensional modeling and the construction of dimensional model, this dissertation proposes three types of constructing consistency dimensionality strategies in data control center of extended independent data marts and the basic approaches of dividing up granularity of dimensional modeling based on time rotation model. Then this dissertation introduced the process and method of implementing the dimensional model of Shenyang superintending inspection decision support system for special equipments. Data transformation is the precondition and guarantee to get high quality data for users. This dissertation discusses carefully the process and method of the data transformation in data marts, and proposes a data transformation model based on dimensionality. After analyzing the multistage and pipeline data transformation strategies of independent data marts based on dimensional model, this dissertation proposes the multistage data transformation algorithm based on dimensional model of independent data marts in heterogeneous environment with XML.Finally, this dissertation studies the producing background of OLAP, the relationships between OLTP and OLAP and the basic analysis methods deeply. As an instance, the design and the implement process of OLAP with the inspective charge dimensional model of the decision support system of Shenyang superintending inspection for special equipment is discussed.更多还原