Research On Web Usage Data Mining Technology And Application

Data Mining is the process that people discover extract, connotative, unknown and valuable information or mode from large database or data warehouse. With Web high-speed development and becomes popular abroadly, a great deal of web log is collected on Web server. These log are recorded every times when the user accesses the website, which are valuable information-wealth. Association rules mined from Web log can provide konwledge about network stream flow and user visit mode in order to improve the performance of web system and optimize the system design of web site, furthermore website can forecast the amount and interests of potential user. It can also instruct the corporation to forecast market abvisably in the field of booming E-business.Web Log Mining is applying Web Mining in analysis of web server log in order to solve problems mentioned above. Based on web log of HanNan Univercity, this paper focused on the following aspects:1) Aanalyze and reserch technology of Web Log Mining;2) Bring forward a multi-path partitioning cluster method, which partions and clusteres according to user access path to improve effiency;3) Bring forward a hierarchy search and iterative method to find frequent set. The efficient method improve the classical algorithm Aprior and reduce times of scanning database to only once when obtain the k-frequent set from k-candidate set;4) Design and realize a Web log mining prototype system, and evaluate it by experiment with web log of HanNan Univercity.5) What is more, this paper discusses how to improve website topylogy with achievement of Web Log Mining.The research of this paper can provide some value of Web Mining development and the technology of improving website topylogy as well as realising adaptive website providing personalization service. It is a feasible way to research and develop Web Log Mining system.