Research And Implementation Of Text Clustering For Personalized News Recommandation System

With the rapid development of the Internet, people have to face mass information everyday. How to get valuable information quickly becomes an urgent problem to be solved. Search engine is not enough to obtain high quality information for every user. In order to deal with this challenge, personalized news recommendation system becomes a hot area of research in recent years.In this thesis, at first the personalized recommendation system of news text clustering related technologies and existing clustering algorithms are studied. According to the special demand of news recommendation system which is based on the content and the topic, we propose an improved text clustering algorithm. Hope that through the use of this kind of scheme to improve the news clustering performance. Then, in view of the news clustering, we designe and realize news cluster system based on the improved algorithm. In order to validate the system availability, accuracy and efficiency, we use this system for mass news data automatic clust, get hierarchic structure of news clustering results. Finally, the results and standards of artificial classification text results are compared. Through the clustering result evaluation, algorithm improvement is proved effect.This thesis is organized as follows:In chapter2, the personalized news recommendation engine is summarized, the performance bottleneck of system is analyzed, and text clustering technology is expected as a solution. In chapter3, text clustering technology is introduced, and several main text clustering algorithms are analyzed. Chapter4, in view of the special needs of news recommend system, puts forward a news clustering scheme in the basis of improved LSH text clustering algorithms.Chapter5introduces the design and implementationof the news clustering system. Chapter6gives and analyzes the testing and experimental results.