Research On Multi-Dimension Query Analysis Algorithm

In recent years, along with the development of database technology and the expanding of database scale, people hope to refine the useful information from existing data to serve for decision maker; therefore, decision support system emerges as the times require. There have been growing interests in the techniques of date warehouse, OLAP and data mining since they strongly support the decision-making. Both as the important applicative techniques of data warehouse, OLAP and data mining have notable differences. OLAP offers user a view of all angles in order to study the data deeply; however, data mining analyzes the collecting data automatically, makes the including ratiocination and then helps the decision maker to adjust the strategy and to reduce the risk, then to gain the correct decision. Association rules'mining is an important topic in data mining. Many research works focus on the relation among item set without concerning the environment (such as time, location etc.). The integration of the OLAP and data mining push the environmental information into the association rules, so that we can obtain the information, which traditional association rules cannot offer. It has very important practical significance. Under this point, this paper studies and implements the multi-dimensional query analysis system, and then, a set of multi-dimensional association rules algorithms base on OLAP technique are brought forward. Main works include: 1) We proceed from designing the system function, then describe the system architecture and explain the every module. Based on the study of designing pattern of multi-dimensional data model, three designation methods are pointed out. 2) Through the deep research of association rules we extended it from intra-dimension and single level to inter-dimension and multi-level. Based on that, we put forward effective algorithms and optimize it for different dataset,besides, we analyzed the complexity of the algorithms and compared the efficiency of the different algorithms for different dataset. 3) Concluded the thesis and described the future work.