Internet Public Opinion Monitoring And Analysis System To Achieve

Internet has continuously become the important place where public opinions are produced and spread,and the public opinions on the Internet are playing a more and more important role in the social lives.In order to enhance the monitoring of Internet, the analysis of Internet pulic opinions is a challenging problem faced by goverments. The Internet Opinions Monitoring and Analyzing System aims to automatically monitor and analyze the huge amount of Internet pubic opinions in real time.The first step is automatic detection and analysis of Internet public opinions hotspots,which can give the government a quick understanding and mastery of current hotspots on the Internet.To deal with the practically applied requirement,we in this paper further promote the current research on the automatic detection and in-depth analysis of pubic opinions hotspots on the Internet,proposed a practical algorithm to automatically detect public opinions hotspots.Considering the hotspot characterstics and people's recognition law,this algorithm introduces several steps including topic sorting,topic combination and adjustment,document elimination and topic description.The experiment shows that the proposed algorithm,in the practical environment,greatly improves the effect and the practicability ofhotspot detection.For topic-focused multi-document keyword extraction,we propsed the POS-tagging based phrase structure method,which effectively utilizes the important global information of many documents,and filters the redundancy information to the best of our ability.The experiments on the Chinese test data demonstrate the validility of our proposed method.A newquery-based tracker building algorithm with new feature selection method is used fortopic tracking.K-means clustering algorithm is sensitive to outliers,so this paper putforward an idea of separating the clustering centroid from the clustering seed toimprove the k-means algorithm and get more veracious and more stable results.News topics are identified from newsstories with topic detection technique and then clustered into specials,Also this modeluses topic tracking technique to track given topics and provides better services.Last,a real application-Internet Public Opinions Monitoring and Analyzing System is introduced and the significance of the automatic detection and analysis of Internet pubic opinions hotspots is demonstrated.