Research Of The User Model In Web Mining

With the rapid expansion of the Internet,a great deal of information has been brought for us. But at the same time, we are also easy to lose the direction in the Internet. The emergence of search engines has solved the problem in some extent, but they can not find the information initiatively that we need. For this reason, now, the personalized services information technology has become the hot areas of research. And the most important of the personalized service is to establish the user interesting model accurately.Firstly, in this paper, focusing on the user model, the key technologies were deeply studied. These key technologies included the classification of modeling techniques, the way to get data sources, the representation of the model, the way to get user's feedback and the means of updating the model. And these key technologies were analyzed and compared.Secondly, based on the above work, a personalized user model was proposed, which oriented single user, combined the user manual modeling and system automatic modeling, combined content mining and behavior mining. Then the paper discussed the generation algorithm and update algorithm of the user model, and technologies including data pretreatment, the weight calculation of the user interesting feature words, the construction of page feature vector, the construction of page behavior vector, and so on. And this paper did some simulation experiments for the user model's generation. Thirdly, based on the user model, a personalized web mining model that oriented single user was designed. Then, the frame and the function of each module of the model were described in detail. And the personalized recommendation algorithm was given.The future work of this paper will focus on optimize the performance of the user model, and apply it in the personalized recommendation service.