The Personalized Service Research Based On Web Log Mining

With the rapid popularization and extensive application of the internet, the amount of the web-based information and the complexity of the site design also shows a sharp growth trend. On the one hand, it is more and more concerned about how quickly and efficiently to find potentially valuable information from a range of the network information. On the other hand, web service providers are also constantly trying to get the interests and hobbies of the users in order to provide them with more targeted services. However, web pages are far more complex than the text document because web is dynamic and unstructured. Web log mining combines the traditional data mining techniques with web technologies to carry out excavation and analysis on the server log, and discover association rules from the vast amounts of information data to address the various issues raised above.Now, focusing on user features and providing personalized service has been a research of hotspot in web technologies. The application of web log mining techniques, combined with Web site content and semantic information is a new trend of the web personalized service research based on web log mining.This paper first outlines the concept of web log mining, the application scope and research contents. On the basis of the current research situation at home and abroad, focusly analyses the implementation process of web log mining system, and gives the data preprocessing and difficulty analysis; of these preprocessing data cleaning, user identification, session identification and path completion are included. Second, web log mining algorithm is studied. A new preprocessing model is proposed, based on analysis of client-side cookies file, to avoid the various web users accessing the websites through the same proxy server can not be identified. User access pattern is mined applying clustering algorithm based on the user browsing behavior and web page clustering algorithm. Associated matrix approach directly processes the site topology and user browsing information, avoiding the complex session identification, and making the process more efficient and more able to tap. Again describes the key technologies for personalized service, and forms a personalized service recommendation system. Through the user access pattern mining before, interested pages of the user are predicted and recommendation results are formed, then these pages the links pointed to are recommended to users by the way of adding the dynamic linking. For different users different website pages are showed to achieve a personalized service. In this paper, the architecture of the system model is provided and the purpose of each module is explained. Finally analyses and summarizes the application and development of personalized service.