Web User Clustering Research Based On The Fuzzy Theory

China's economic and information technology's rapidly development, the website, portable devices, networks have been deeply rooted in people's life; people have more change to communicate with unknown people. In education and science research field, computers and networks are play an more and more vital role for assisting teachers and students'study and research. In the area of web-based education, what is effect way to find the user's real intention? And how to classify them exactly to different clusters in order to improve website's service and find out new business opportunity has been one problem.This thesis majorly focused on the website user's classification that based on the technical of fuzzy math, data mining and cluster analysis. Its intention is to find the user's real intention and his characteristic, and then classify them into different clusters. After that, the website can recommend special resources to different cluster users, and enlarge its income. On this background, this paper focuses on the research of web user clustering. With China Education TV GuoShi website as the research object, combined with the field of fuzzy clustering and equality matrix, we bring forward new method to measure and adjust the user interest and attitude, which can provide more accurate source data to clustering system. Achievement as following shows:1. Designed all-purpose, flexible clustering system architecture for GuoShi website. This system can be used to cluster different objects, such as user and resource. It is also very easy to replace and add new clustering algorithms for this clustering system. Its performance and practicability has been approved in the real world.2. Provide a method to measure and adjust the use's trait which changes with time. Based on the interaction between the user and the website, combined with oblivion curve theory, this method can forecast its changes, which is very useful for the clustering system.3. Combined with fuzzy similarity transfer closer, bring forward a method to figure out the number of initial clusters'count. This method can avoid the arbitrary, also it will improve the result's veracity.