Study Session Session-based Site Recommendation System

Internet provides us with a wealth of information, but also bring people to "spread information" concerns. The traditional site to the people the impression that most of the information listed, although rich in content, but the people is difficult to find the content of their needs. Although sometimes we can through the search engine to find information, but the e-commerce sites, it can not reflect its business services personalized features, e-commerce in today's highly competitive era, how they need to provide users with the services appear to be particularly Important. With data mining technology, people can from the historical data to find the behaviour of visitors, and provide intelligent and personalized service to visitors. Using Web data mining results, we can optimize the structure of the site, improve the system performance of visits, through the historical data found valuable business information, and for visit behavior characteristics of different users, providing personalized information pages or commodities recommended Services, so as to avoaid to find information in the blindness for people.Based on the analysis of Web site data storage characteristics, This article has elaborated on the concept, process and methods, and other related technologies of Web data mining, for the shortcoming of the Web log past treatment to obtain visitor behavior, based on the proposed Session conversation server for data acquisition way to the data source, through set up to track the current Session conversations online access in server, and recorded the sccess data to the data warehouse model which prior to establish for analysis. compared to the method of Web log processing,it can be greatly simplified Web data acquisition and pre-processing program, increase in user identification and recognition of the accuracy at the conversation, it can be realized for users to access real-time analysis of the situation.This article studys in-depth the site personalized information system recommended by the structure, function and algorithm principle. The Personalized recommendation system is composed of the online recommendation and offline analysis, for us to the large number of data analysis of the current users to conduct real-time online processing. The visit length of pages which user visited is looked on as the measure of interest in this system, through the collaborative filtering algorithms and cluster analysis technology, will divide into groups for different characteristics of the users, when new users access to Web sites, the system will recommend According to the user's previous visit to conduct their classification, the user owned by the user cluster, and then recommend to those who have the same characteristics of cluster users visit the page-interest. This paper describes the realization process of the personalized information recommendation in theory.