Design And Implementation Of Multidimensional Data Analysis Platform Based On MDA

With the development of information technology and the popularization of MIS (Management Information System), the business processing database of each industry has accumulated a large amount of rich data. People hope that they can explore rules and find knowledge from these vast amounts of business data to support management decision. However, the ordinary business processing systems tend to focus on real-time business processing. The capability of data analysis of these systems is limited, the form of showing data is single, and these systems are hard to upgrade. It is difficult to meet the needs of analysising data in depth and multi-dimension.Therefore, it becomes an important research topic that how to develop a system which can support multi-dimensional data analysis with low cost rapidly. Against the problems and shortcomings existed in the data analysis of the ordinary business systems, this paper presents a multi-dimensional data analysis platform based on MDA. The platform takes full advantage of the idea of MDA and provides an efficient solution for building multi-dimensional data analysis system through the researches on construction of the data warehouse, data integration, statistical analysis, data visualization and so on. The platform is based on the technology of J2EE, and uses the three-layer-model architecture. By using the framework, developers can complete functional development on data analysis and data visualization quickly.In the implementation process of the platform, by analyzing the process of the implementation of Business Intelligence, this paper extracts various models in the multi-dimensional data analysis to point out the direction for the model-driven software development. This paper uses XML as the language for defining model, and designs and manages the meta model depending on CWM specification.By analyzing the relationship between fact tables and dimension tables in the data warehouse, this paper clear the way of building data warehouse. And this paper completes the multi-dimensional data modeling based on the star schema, forms a model description standard, and defines a platform independent multi-dimensional data model. And by using multi-dimensional data model as drivers, this paper designs model parsing engine and outputs for a platform specific model instance, thereby it builds a data warehouse and solves the non-standard data format problem in the ordinary business processing systems. The article also analyzes the process of data integration of data warehouse, abstracts model of the ETL process, defines the ETL model, defines the ETL platform-independent model with XML as the carrier and obtains the corresponding SQL and Java code to implement data extraction, transform, and load through the conversion of the model, so we can put the data scattered in different business database together into the data warehouse.To implement the multidimensional analysis of data, this paper researches methods and processes of the OLAP, and defines OLAP model which can be converted to database SQL query to implement query and analysis of multi-dimensional data. At the same time, this article implements the method for data visualization display which allows users observer and analyze data intuitively by using Flex-based technology.This paper integrates the idea of MDA and the development of multi-dimensional data analysis system closely, has done the exploratory study and provides an efficient solution for the development of multi-dimensional data analysis system. This solution lowers the cost of development, improves the development efficiency, commonality and reusability at the same time. This make the subject in this paper has a certain theoretical significance and application value.