Research And Implementation On User Interests Modeling For Personalized Information Retrieval

Information rescources on internet grow explosively, leading to the problem of"information overload but resouces to the crowds". It brings great challenge to information retrieval. It is passive to accept request of users for traditional information retrieval, instead of proactive to perceive users'information needs. Traditional information retrieval has been very difficult for users to meet the rising complex and differentiated information needs. So, users prefer personalized information retrieval more and more. This paper research user interests modeling that is basis and core of personalized information retrieval.The content of this paper is as follows:(1) Because it is not accurate or comprehensive for VSM to express user interests, and the VSM do not consider multiformity of user interests. This paper poses that hierarchical VSM express user interests. The experiment proves that it can be accurate and comprehensive for hierarchical VSM to express user interests.(2) Because page interest degree based on page browsing time and page browsing times do not consider the size of page, this paper computes the page interest degree based on page browsing speed. The experiment proves that the result computed based on this method is close to fact value.(3) Because of hierarchy and extension of user interest model, xml is used to save user interests and inverse index mode is used to map the user and corresponding xml saving his interests. It can be quickly to find xml saving a user's interests according to the user's name.(4) Posting an improved method to build user interest model, design and implement all modules of user interest modeling, and apply user interest model to nutch search engine, and implement the personalized search engine. The experiment proves that the personalized search engine is better on accuracy compared to the traditional search engine.