Studies On Hot Topics Discovery Of Public Opinion On The Internet

With the rapid development of Internet and the explosive growth of internet information,the network has become an important way for people to release and get information.Facing the vast internet information, how to get useful information from it has become theurgent problem in today’s society. In this context, identification and analysis of heatnetworks hot topics have become more and more important, and they will become hotspots for current researchers.In this paper, the Internet news reports hot topics were analyzed using topic automaticdiscovery and topic heat analysis technique. The main research content is as follows:(1) This paper puts forward the algorithm of feature extraction based on wordfrequency. The algorithm for extracting nouns and verbs in the news text that has beenmarked as a feature, use matching rules. Experiments show that the feature extraction toimprove the precision and speed reduce the feature vector dimension, so as to solve theproblem of the vector explosion.(2) Through the analysis of the structure of the network news reports, in featureweighting computation, word frequency statistics appear in the title and contentrespectively, and will feature multiplied by a weighting factor that appears in the title, thusincreasing the weighting feature may become hot. Experiments have shown this methodimproves the hot topic detection.(3) A network topic found algorithm was put forward. Through the traditionalclustering algorithm of comparison and analysis, overcome its shortcomings, puts forwardthe strategy based on second floor of the clustering algorithm. It would first of all dailynews reports on local clustering, and then in the past with the topic of merge clustering.Experiments show that the algorithm reduces the residual rate, error detection rate andloss cost, so as to improve the topic discovery ability. Experiments show that thealgorithm reduces the residual rate, error detection rate and loss cost, so as to improve thetopic discovery ability.(4) Puts forward the network news hot topics found model. Analysis of thecharacteristics of network news, constructing the network news hot topics measure. Putforward the media attention and user attention network news hot topics found model,based on the measure, experiments show that this model can automatic find a time website hot topics, and the comprehensive analysis of network news topic heat condition.(5) Inspired by the stock market curve, the introduction of topic index and topicdevelopment curve. The model through the analysis of site each news topic heat, forming a hot topic development curves. Experiments show that topic development trend and theactual match.