The Research Of Web Usage Mining Based On Clustering

With the popularity of Internet and the explosion of WWW, Web usage information is becoming the largest information warehouse. As a result, the conflict between the limited human attention and the unlimited information is becoming more and more serious. Luckily, Web Usage Mining is found to be an efective method to solve this problem. It can be used to find out users' normal browsing paterns and give significant advises for web site supervisors by extracts interesting paterns from the original Web usage information. By this means, Web Usage Mining is significant for Web site design and management, business and marketing decision support and users attraction.This paper first focuses on the theoretical and technical foundation related to web usage mining, such as clustering and association rules. It has formed the systematic framework for web usage mining and illustrated the approach to web usage mining in terms of data preprocessing, discovering user visiting pattern, web clustering and characteristic rules, etc.In the field of data preprocessing, this paper systematically studies the processes of data cleaning, user identification, session identification, path completion and transaction identification. We have designed some related algorithms.In the field of discovering user visiting pattern, this paper researches the theory and method for user visiting pattern. On the basis of analyzing preference, we have proposed the method of finding frequent browsing path based on page level and degree of liking. Finally discussed the technique of finding frequent visiting path considering the support, browsing time and page level.In the field of web clustering, the paper systematically studies the methods of determining the likeness of clustering objects. Considering the features of the web, the paper also studies the methods of web page clustering based on link distance between pages. In the field of web user clustering, we explore a clustering approach considering interaction and influence among clustering objects.In a word, the paper systematically discusses the process of web usage mining, theory and method for mining implicit knowledge from huge and semistructured web data, to provide a broad prospect for the development of web usage mining.