Internet Information Filtering Agent And Achieve

With the rapid development of Internet, the information on it increases in exponent. It has so huge contents and so many sorts that it may be the largest library in the world. How to search out what users interest on decides whether we can utilize the huge information resources and it is the research goal of our article. Here we use several machine-learning methods and the agent technique to develop a new filter Agent which has intelligence, and efficiency. The major research contains two parts: an information filter system and an interest learning system. The information filter system contain: Chinese words take-out module, and information filters module. In Chinese words take-out module we adopt an approach using the characters of Chinese to pre-process the downloaded web pages, and an approach of max-matching and frequency statistics. We use a keyword vector method in information filter and improve the method as follows: firstly, assigning different weight to the keywords on different tag; secondly, an user can modify weights of these keywords. By those methods we can get the articles that can reflect the user' s interest better.In interest-learn system, we adopt a vector space approach and the id3 algorithm. And we adopt learning-methods based on user' s feedbacks, observe user act in the background and provide the function of severing user' s interest initiatively.This article adopts several machine-learning techniques including machine-learning method based on user' s feedbacks,machine-learning method based on observing, inductive method based on id3 algorithm, heuristic learning method, knowledge base technique. The system has good intelligence and agency.