Design And Implementation Of Public Opinion Analysis System Based On BBS Topic Comments

With the development and popularization of the Internet, the Internet has become the"fourth media" after newspapers, radio and television. More and more people are willing toshare information on the Internet, express their views and opinions about some topics.Forums and microblogging has become an important communication carriers of socialinformation. However, because of the easy access to the Internet and openness of Internet,at a certain level,making the network become a breeding ground for the proliferation ofrumors, arouse social contradictions, affect social stability. How to effectivel y monitornetwork security public opinion is very important.With topics and the associated information of BBS forum as the object of the research,the text in natural language processing emotional tendencies as the basis and research onheat trend for the topic, design and implement a network public opinion analysis systemwith fuinction of forecasting heat trend of topic.The main contents and innovations are as follows:(1) Point out the importance and necessity of monitoring network public opinion,introduce research status of related fields, and summarize technologies and theories todesign and implement this system. This paper also raises a general solutions framework tomonitor network public opinion, provide practical ideas for the other researchers.(2) Design and implement the complete system. This system includes public opinionsentiment analysis module based on text and public opinion monitoring module, with thebasis of HCI module, database and index files module. In this paper, we introduce theprocess of implementation from data acquisition, database design, emotional analysis, heattrend forecasting of topics and interactive interface for detail. This system can get data fromwebsite, analysize the emotion of data and forecast the heat trend of topic based onanalysized data.(3) Fixed existing heat evaluation methods, proposed a new heat trend forecastingmodel, and implement this model in system. The existing heat evaluation methods onlyconsiders the number of clicks and replies, and ignore the problem that the Views tend canimpact the topics’ heat. Such predictive models for topics’ heat trend is not consistent withthe actual situation. In this paper, we present a hot topic prediction methods that fusion ofview of the tendency factor. We can predict the heat trend of topic using Gama model.Experimental results show that the results of the improved method to get closer to the actual topic trends.