Research Of Online Analytical Mining Technology Based On Data Warehouse

Nowdays, the information is expanding in our society. Analysis and concern about information processing are gradually improving with each passing day. Therefore, the efficient data analyzing technologie has became a hot research focuses in current information technology field. As the important technologies of Decision Support, OLAP and data mining own their advantages and shortcomings individually while completing decision analytical tasks.If user adopts OLAP to analyse issues, he has to suppose what will happen. Then the OLAP will execute a dimensional analysis for the supposition and analysis result will prove supposition is right or not. So it is very susceptible to be affected by users'subjective, which will affect the accuracy of analysis results. DM will execute mining task automatically after algorithm is determined and user cannot intervene the whole process. So this process is blind. How to integrate OLAP with DM to improve efficient and valuable of data analyzing technology has become a hotpot problem. Therefore, J.W.Han brought forward the concept of the OLAM in 1997. It had combined the advantages of the above and offered the method of the interactive data mining based on the OLAP technology.The efficient data mining algorithm is the core of the OLAM mining mechanism. This paper choiced a widely-used association rules algorithm to analyse, and carried out in-depth research on multidimension and multilayer of the association rules. Based on the research of OLAM system structure, this paper proposes the structure of multilayer and multidimension association rules system which based on the data cube technology.In this thesis, on the base of research the relevant concept of association rule mining, OLAP and data cube, we summarize the existing multidimensional and multilayer association rule mining system based on data cube, and improve this algorithm in three aspects. First, in order to mine more interesting rules to users, we introduce the OLAP operations into the mining process to adjust dimensional level of the cube dynamically. Second, in the process of creating frequent itemsets, this paper uses the Hash technology to filtrate the candidate frequent itemsets. It improves the efficient of algorithm. At last, this paper introduces an association rules producing algorithm which is based on back piece. It could reduce redundant rules, and improve the interest of rules.At the end of this paper, we developed the main algorithm to test the improved methods better.