MDA在数据仓库维度建模中的应用研究

【作者】 刘淑芝；

【导师】 吴海涛；

【作者基本信息】 上海师范大学 ， 计算机应用技术， 2007， 硕士

【摘要】 数据仓库的设计涉及到多个层面,如ETL过程、可操作数据源、数据仓库存储等。每个层面都会有不同的建模方法,这些方法往往只是针对单个方面提出的,这样建模制品就会缺乏一致性和可复用性,设计者很难从一个统一的、整体的角度对数据仓库进行设计。模型驱动架构(MDA)是一种用于软件开发的标准框架,着重用模型描述软件开发的整个生命周期。把MDA应用于数据仓库开发中,能很好地解决设计中的整体性和统一性问题。维度建模是数据仓库存储设计的基础,本文针对维度建模中存在的一些问题,如无法从多维概念模型自动转换成多维逻辑模型,建模制品可复用性不高、且不利于数据仓库元数据的交换或共享等,提出了一种把MDA和维度建模结合起来的方法——模型驱动的维度建模(Model Driven Demensional Modeling MDDM)。本文用扩展的UML进行多维概念的建模,然后通过形式化方法——查询/视图/转换(QVT)清晰地定义出多维概念模型到逻辑模型的转换,使其具有可重用性,并实现了多维概念模型到逻辑模型的自动转换,转换结果遵从公共数据仓库元模型(CWM)标准。这样既有利于数据仓库中不同平台和工具间的元数据的交换和集成,又降低了设计者对CWM建模的难度,同时提高了建模制品的可复用性。更多还原

【Abstract】 Several aspects should be considered in the design of data warehouse (DW) System, such as ETL process, operational data source, DW repository, etc. There are different methods for modeling in the processes of designing different aspects of DW.However they are all targeted at particular aspects of DW design, as a consequence, the artifacts of modeling are lacking in coherence and model reusability, designers find it hard to design data warehouse in an integrated and standard manner. While Model Driven Architecture (MDA) is a standard framework for the software development, it can describe the whole life cycle by means of models. The combination of MDA and data warehouse can help to solve the problem of lacking in coherence and reusibility.Dimensional modeling is the foundation for the design of DW repository. This paper proposes a method which called Model Driven Dimensional Modeling (MDDM). It applies MDA to dimensional modeling so as to solve some of the problems in the dimensional modeling. For instance, 1) some methods can not shift from multidimensional conceptual model to multidimensional logical model automatically. 2) The reusibility of models is not very promising. 3) The artifacts of the modeling can not facilitate the interchange and share of metadata.With the MDDM method, a multidimensional conceptual model is built using extended UML.Therefore it can provide the designer a standard and integrated way to design DW repository. The transformation from conceptual models to logical ones is defined clearly by the formal way Query/View/Transform (QVT). As a result the transformation is reusable. The conceptual model can be automatically transformed into a logical model which is compliant with Common Warehouse Model (CWM). Therefore the complexity of CWM modeling is decreased, and the reusibility of the modeling artifact is improved. It also helps the interchange and integration of the metadata in data warehouse.更多还原