Research On Interest Learning Methods In Personalized Search Engine

With the rapid development of Internet technology, the network can provide people more and more information. Search engine has been widely used in many fields, which is treated as a tool that people can get information on World Wide Web.In order to provide personalized search service for users,personalized search engine comes forth. Because of many researchers contribution,people have made great progress in personalized search engine.This paper points out the shortage of current search engine and user's requirements of personalized search, does some research on interest learning methods and its technology, and designs a interest learning profile.The main tasks of the paper are as follows:Firstly, this paper designs a interest learning profile.The profile designed on the paper gets user's interests from the pages that he has visited before,puts interest words into user's interest database, and uses the short interest and long interest to describe the user's interest characters. In order to reflect user's interests changing in time, this paper updates user's interests using the algorithm based on Genetic Algorithm. In order to evaluate users' interests comprehensively, this paper introduces weights, files' inreresting degree and forgetting mechanism. And file's interesting degree is based on Genetic Algorithm and BP neural network,it can reflect user's interests more accurately.Secondly, this paper researchs and implements the interest learning methods.This paper researchs the methods of interests learning,uses direct learning, history learning and real-time learning.This paper uses interest learning and the technology of personalized search engine, designs and implements the system of interest learning.At last,experiments are conducted to verity the efficacy and exactly of the interest learning profile designed above.The contents of the experiment include twoparts:generate interests and update interests.