Research On Personalized Technique Of Data Mining In Information Retrieval

With the rapid development of Internet, web information are expanding, People have to spend a lot of time to search the information they need. Most of the search engines can not satisfy the requirements of users'. How to mining personalized resources for different users from a lot of network information has come to be a hot research. This paper designs and carries out an intelligent information retrieval system, it does a deep study on personalized information mining module which is the core module of the system.This paper firstly analyzes the principle and the architecture of meta-search engine, discusses the application of Agent on information retrieval. Then it studies the key technologies about Web text mining and analyzes the automatic text classification technology. The paper puts forward an improved text classification method based on neural network and does a feasible test later. Then the paper analyzes the document vector space model. According to the characters of web labels, a weighted method which is classified by labels is given. Moreover, aiming at the characteristics of user's interest and access frequency, this paper adopts a series of steps that including text pre-disposal, user identification and session identification to build a dynamic three-tier interest model. The paper uses the model to filter the searched information which satisfies the users'personalized needs very well.