The Research And Application Of Web Usage Mining

With the development of computer technology especially Internet, the usage on various web servers are also expanding rapidly. Although the traditional database techniques and data mining techniques have developed very rapidly, it can not be applied directly to the web usage mining Because most of web usage are incomplete,redundant and inaccurate, some of the traditional techniques can be applicable to web usage mining only if they are based on the preprocessing of web usage.Web usage mining aims at picking up usage patterns from web usage through effective web usage mining. We can get valuable information we need such as usage patterns by analyzing the behavior,frequency and content of users'access on web. This information can contribute to the improvement of web design and, what's more important; provide personalized services to users through the analysis and understanding of users'characterization.This paper has conducted in-depth study on every procedure of web usage mining. Firstly, the process of web usage preprocessing is introduced based on the summarization of the inland and oversee study on the preprocessing of web usage. Secondly, a new Association Rules Algorithm based on Interestingness-selection is introduced. This algorithm has analyzed not only web usage but also other valuable information such as web page content,the topology of web site and web page size. This algorithm has also applied interestingness to measure the quality of association rules, which helps to greatly improve the interestingness of resulting rule and then produces the frequent access web pages set of high interestingness. Finally, the method of online personalization recommendation, which is based on the new algorithm, is put forward. This method can produce users'relevant web page recommendation set by using Association Rules Algorithm and Mode Matching Algorithm, and the recommendation engine can provide online personalized recommendation to users.