The Research On The Use Access Pattern Mining Based On Web

Along with the rapid development of network scale and the swelling requirement of user, web has been applied to all kinds of fields. More and more organizations, companies and people publish or search information on the internet. Internet not only is full of text, audio, video and other multimedia information but also includes linking between pages, use access patterns while people surf on the net. People suggest to discovery useful knowledge hided in the huge information by data mining, but the features of internet is different and complicated, people combine data mining technology and the web to create the web technology.The web log file records massive path information of users, when they browse the site. It is useful to help web site designers to obtain users' habits on the base of analysis of these information. And also can optimize the structure of web site and improve the web server's performance.The thesis begins with the web log mining system and search large associated data from domestic and abroad. In the aspect of how to mining web log with high availability and how to discovery knowledge, author has made more deeply research. The main research work is mining use access patterns out from the web log, just the interest patterns. And to recommend the pages that the users need.First, this thesis introduces some theory knowledge, such as the background knowledge of data mining and web data mining, the conception of web data mining and the classification of web data mining. It emphasizes on the knowledge of web usage mining (web log mining) .Then, this thesis introduces the web log mining stage and disserts the data-preprocessing stage in the web log mining aiming at special log data format. At the stage of pattern mining, this thesis emphasis on how to find the users' interest patterns and puts forward a kind of ways to mining the use access patterns based on the user visitor act. This thesis proposes a new parameters-frequency and interest degree to traditional algorithm, and tests it in theory and practice that improved the algorithm really has its advantage.At last, this thesis discusses how to apply the use access patterns to the personality recommendation. The principle is to recommend the pages that the users really need then provide the characteristic service for the users according the use access pattern mined out from the web log. It can get accurate results quickly.