Research On Data Warehouse System Based On MDA

Data warehouses play a central role in current Decision Support Systems because they provide crucial business information to improve strategic decision-making processes. However, building a DW is still a challenging and complex task because it consists of several interrelated components with different functions, different design pitfalls, and different technologies. Different methods and approaches have been proposed for the design of each DW layer.Nevertheless, these methods do not deal with the design of the whole DW in an integrated modeling framework. Model Driven Architecture (MDA) framework is the next-generation solutions to the problem of distributed, heterogeneous systems. It contains platform-independent model (PIM), platform-related model (PSM) and the code (CODE).Multidimensional data modeling is the date organizing form in data warehouse and it is also a key factor in the data warehouse application. On one hand, the MD is close to the way of thinking of data analyzers and, therefore, helps users understand data better. On the other hand, the MD model supports performance improvement as its simple structure allows us to predict end users'intensions.This thesis mainly describes how to build the different MDA models for the DW repository by using an extension of the Unified Modeling Language (UML) and the Common Warehouse Metamodel (CWM). Transformation between PIM and PSM is also clearly and formally established by using the Query/View/Transformation (QVT) language, and implement the automatic transformation between PIM and PSM by the open source tool of mediniQVT. At the same time, this thesis has set up and realized the transformation of PSM to SQL code under EMF framework, Thus realized the model driven development based on model. Finally, a case study of the subject of purchased price analysis in an iron company is provided to exemplify the benefits of our MDA framework. Practice has shown that not only the Development efficiency of developing this system using this model has improved , but also, because this model is established based on the international standards Modeling Language (UML), so the model is more standardized, and is very good at achieving share between different models, at the same time, it is conducive to reuse and maintain.