Design And Implementation Of Topic Driven Network Public Opinion Monitoring System

With the development and popularization of the Internet,the Internet has become a very important way for the people to obtain and release information.Netizens can comment on an event or some policy free through the Internet,this makes public opinion acquire the expression of an unprecedented space with the help of Internet communication channels and realize the rapid and extensive expansion of transmission area.The Internet has become the position of public opinion dissemination of information and the main battlefield of network ideological struggle of public opinion.Facing the surge of network public opinion,how to improve the occasion,correctly dealing with the public opinion,to reduce negative impact of the network of rumors in the public opinion,for effective monitoring of network public opinion,analysis and control of government departments at all levels and enterprises has become an important issue.Through the research of network public opinion,this paper designed and successfully implemented news topic driven network public opinion monitoring system.News topic driven network public opinion monitoring system refers to according to the specified topic of users or relevant departments from the site search and collect the relevant news reports,and with the news reports implement topic detection and sentiment analysis,thus provides decision support for the user or the relevant departments.According to the system,this paper conducted the following research work.It used LAD model to implement topic detection and built a basic emotional lexicon,negative word dictionary,adverbs of degree dictionary,etc.,with the method based on dictionary and rules implemented sentiment analysis.It implemented topic driven network public opinion monitoring system.The system has web page data acquisition,data preprocessing and public opinion analysis the three big functions.with technologies such as meta search,web crawler from nearly 50 websites at home and abroad,mainly has phoenix,netease,sina,news agency and other domestic websites,as well as,the New York times,the Times,Reuters foreign news websites searched web data.Again by word segmentation,text representation and feature extraction technology preprocessed information extracted from web pages.With the LDA model implemented topic detection and used dictionaries and rules of sentiment analysis method to implement sentiment analysis for the public opinion information.This topic driven network public opinion monitoring system has repeated test run in the real network.The experimental results showed that the system can all-sides effective find information for network public opinion at home and abroad and implement topic detection and sentiment analysis.