Research on clustering analysis of Internet public opinion

In order to enhance the real-time performance of Internet public opinion recognizing and early warning, and improve the accuracy of the analysis of Internet public opinion for hot spots, similarity analysis methods of Internet public opinion are put forward. Firstly, web crawler technology is introduced for obtaining accurate and comprehensive public opinion. Secondly, propose similarity algorithms from the aspects of known and unknown of the subject. At the same time, considering the uncertainty and fuzzy of Internet public opinion, the concept of information entropy is introduced, and present a similarity analysis approach of Internet public opinion based on information entropy, and can cluster and identify hot spots and crisis events of Internet public opinion. Experimental results show that the proposed methods can quickly obtain the Internet public opinion, and has high accuracy rate of clustering, which provide an important technical support for Internet public opinion monitoring and recognizing.