Research And Implementation Of A Recommendation System Based On Web Log Mining

With the rapid development of the Internet applications, the information of Internet prompts increases. While the huge amount of information on the Internet makes it harder for people find what they acquire. Such a phenomenon could be called information overload. At the same time, the wide distribution of information on the Internet makes the users find interested one more difficult that is so-called information astray. Most search engines presently have not yet solve problems including information overload and information astray effectively owing to insufficient in positive characteristic and lack for taking users' interest into consideration.Web log mining is major technology and tools to study Web browser of users to understand the user's interest is an important part of improving the quality of services and the structural design of Web site. Through analysis and research the law of users access situation, can identify the potential customers of e-commerce and enhance the quality of server and improve the structure and performance [1,2] of Web server system. An important research direction of Web log mining technology is the Web users cluster and pages cluster, through users process and research usage information of the web site -Web log file, received a user groups and users interested in the site's URL with interested of similar visit, which can determine and adjust the structure of the site and provide personalized service [3,4]. But the present study there are a number of shortcomings, at first in cluster of similar measure of cluster, simply to use browse time or the number of visits to measure, for this complex case of Web site, the cluster is not accurate. Moreover, they are using the traditional clustering technology, which is each object to be strictly divided into a category, can not handle overlap problem between the categories.In this dissertation, at first introduced data pretreatment process of Web log mining, including data clean, user identification, conversation recognition, path added and services recognition. And analysis the purpose and methods of each steps, and gives the algorithm of each steps. Then, from users clustering, web pages clustering and frequent access path in three respects to consider Web browser pattern. And give some relevant definitions. The vector and fuzzy sets algorithms based on the original clustering algorithm are discovered, for cluster the users and web pages effectively, and produce frequent access path, so as to provide for users personalized recommendations. Finally, to achieve a recommendation system based on Web log mining.