【作者】 李智；

【导师】 王大玲；

【作者基本信息】 东北大学 ， 计算机软件与理论， 2008， 硕士

【摘要】 大规模、跨领域数据仓库的出现,使数据仓库中的数据量日益增多,数据模式日趋复杂,清洗过程设计更加繁琐。这些变化使数据清洗软件的开发出现了很多新的需求,如动态构建、频繁的修改以及更多的用户交互等。因此,如何设计能够可重用的数据清洗过程以满足数据清洗软件新的需求,成为数据清洗软件设计和开发人员的新挑战。为解决上述问题,本文深入地研究了数据清洗的逻辑模型以及构件化物理实现,完成了如下两部分工作：一是提出数据清洗过流程模型DCPM (Data Cleansing Process Model),并证明基于该模型使用工作流网建模技术对数据清洗过程进行建模的可行性。这使得设计数据清洗过程可以遵照统一的模型,并使用成熟的建模技术进行建模。该模型描述了数据清洗流程中内部各个元素,元素之间的关系和元素的属性。通过使用统一的模型对数据清洗流程进行建模,显著地提高了清洗流程的可复用性。二是分析了目前开发数据清洗软件所出现的新需求和已有的开发手段的不足,提出并实现了基于构件的数据清洗框架C+ADC (Component-extended Agile Data Cleaning),包括运行时平台和框架服务构件集。基于该框架,能够轻松、灵活地构建基于扩展构件的数据清洗应用,显著地降低开发成本和周期。定义了集成与该框架的构件模型和数据清洗流程业务空间向构件空间映射策略,以更好地实现数据清洗应用。实际的开发案例证明：使用DCPM模型建模数据清洗流程并基于C+ADC框架进行数据清洗应用开发,能够快速地构建基于构件的灵活的、可扩展的数据清洗应用软件。更多还原

【Abstract】 The appearance of large-scale and interdisciplinary data warehouse brings the increasing data quantity in each data warehouse and more complex data mode, the design of cleansing process also has been more complicated. These changes result in many new requirements in the development of data cleansing software, such as dynamic creation, frequent amendment and more user interaction. Thus, how to design a reusable data cleansing process for meeting the new requirements has been a new challenge to the designers and developers of data cleansing software.In this thesis, the logical model and physic implementation based on component of data cleaning in-depth are studied for solving above problem. The main work is done in two parts as follows:In first part, DCPM (Data Cleansing Process Model) is proposed and the feasibility of modeling data cleansing process by workflow modeling technology is proven, which enable data cleaning process comply with agreed module by mature modeling technique to model. This module describes the attribute and relation among the inner element of data cleansing process flow. Modeling data cleansing process based on the uniform module increases reusability of the process.In second part, through analyzing the new requirements in present design and development of data cleaning software as well as the deficiencies in development method, C+ADC (Component-extended Agile Data Cleaning) is proposed and implemented including a Run-Time platform and a framework service component set. As it is easy and flexible for this frame to construct data cleaning application based on extended component, it reduces the cost and cycle of development a lot. Moreover, the component module integrated C+ADC framework as well as the map strategy from the business space of data cleaning process to the component is defined in order to implement the data cleaning application preferably. A practical development case has proven that development of data cleansing application based on DCPM (Data Cleansing Process Model) and C+ADC (Component-extended Agile Data Cleaning), can construct quickly a flexible and extendable component-based data cleansing application software.更多还原