Design And Implementation Of News Recommendation System Based On Micro-Blog Users

Weibo is the micro-blog, in which people communicate each other at anytime and anywhere. The development of the Internet has made Weibo become a more and more important part of people’s life. Micro-blog contains a large number of information, and the study on micro-blog users currently has become a hot area of research. At the same time, the Internet has a huge amount of news every day, but the personalized news recommendation system can help users easily find news of interest. Considering the large number of users and huge size of information from Weibo, it is significant to design a news recommendation system based on micro-blog users practically.In this thesis, a personalized news recommendation system is designed and implemented by analyzing and studying user’s Weibo information to predict user’s interest. By using this system, users can accurately obtain news of their own interest.This thesis firstly introduces the research background and significance of the subject, and analyzes the current research status at home and abroad. Next, the thesis carries out the demand analysis and the outline design of the whole system. The system is divided into three modules, i.e., data acquisition, interest prediction and news recommendation. And then, the detailed design and the implementation of each module are completed. The data acquisition module is mainly for micro-blog data collection, and the ICE (Internet Communications Engine) middleware technology is adopted in this thesis to implement the distributed crawler which is used to collect micro-blog data. The interest prediction module is used to analyze user’s micro-blog text, so as to get the user’s interest. In this module, some technologies are applied, that is, the Chinese segmentation technology is used to preprocess micro-blog text, the TextRank algorithm is applied to extract keywords, and the word vector is adopted to represent feature representation, and finally a logistic regression model is used to predict user’s interest. The news recommendation module accomplishes personalized news recommendation through user’s interest, and the news RSS (Really Simple Syndication) source is used for news of acquisition, also the content-based recommendation algorithm is applied to recommend news.Finally, experiment and analysis are conducted for each module. The experimental results show that the system can accurately obtain user interest, and carry out corresponding news recommendation based on the interest.