Application Research On Data Warehouse And Knowledge Discovery Based On Complicated College Research Information

With the coming of knowledge society, scientific research level becomes an important indicator to measure the comprehensive strength of colleges. In order to find out problems in research management, managers are anxious to make useful of the historical research data to obtain decision support information. As the historical research data is transaction oriented and can not provide needed information directly to managers, so it is necessary to build data warehouse and implement data analysis and data mining. On-line analytical processing enables managers to observe and analyze the data they are intested in from different angles and levels. Data mining can find the conceal relationship among data. These technologies can be used to obtain valuable information to support management decision.By studying data organization structure of data warehouse, this paper proposes a data organization and analysis structure model of scientific research project, so information at different levels can be obtained. First, the historical research data is collected as the first level data in this model. Based on the analysis of original data and management requirements, a star schema of research project has been designed and the according data extraction, data cleaning and data transformation have been implemented. The processed detail data is stored in relational database SQL Server 2000 and becomes the second level data. Then multi-dimensional data cube has been created using SQL Server 2000 Analysis Services as the tool of OLAP, and data slicing and drilling have be done on the data cube. The integrated data in the cube becomes the third level data. When the data warehouse is created, two levels of data mining of project and the collaboration among teachers have been done based on k-means clustering algorithm using SPSS17.0. Finally, the final results of data mining and integrated data from cube are analyzed and displayed by tables and graphic toolsThe results of analysis shows that in the year after 2007, there are less high level longitudinal projects and main longitudinal funds come from provincial level projects and city level projects. Although the total number of projects is increasing, the main increasing projects are below 200,000RMB. Senior titles are the main work force of research work, but they seldom collaborate with each other. Most projects below 500,000RMB are accomplished only by one senior title or one vice-senior title. In projects above 500,000RMB, the number of projects attended by one senior title and some vice-senior titles increases. However, in 2008 and 2009, teachers seldom collaborate with each other in longitudinal projects above one million RMB. The research in this paper provides different levels information to managers and is meaningful for them to make effective scientific research policy to guide the action of research of teachers.