Internet Public Opinion Hotspot Discovery Technology Research

With the development of information technology, the Internet also gradually become the main position and source of network public opinion. Therefore, to accurately judge the accuracy and security of the information from the Internet content, so as to timely and accurate grasp of the direction of the development of the Internet public opinion is very important. But due to the news on the Internet is a very typical unstructured information, and news content on the network also has the characteristics of no limit, so as to determine the content of the Internet news also will be relatively complicated.In order to ensure the safety of Internet information content of the network information content posted on the Internet is very necessary for the orderly management, through the computer automatically to the network public opinion information are summarized and finishing, to establish the comprehensive supervision, effective quick search of network public opinion to detect early warning system of the problem.Corpus is mainly from the network news public opinion analysis of this paper, the main research how to effectively and quickly in the network public opinion analysis found that the network public opinion hot topic. First in this paper, we study the hot topic to find related technologies, such as:Chinese label, the choice of key words, text clustering, etc. Secondly, set up the system, found on the hot issues of network public opinion system mainly includes:public opinion information collection, pretreatment, such as the discovery of a hot topic, and then, the advantages of using traditional K-Means algorithm and headlines the importance of word frequency found in the hot topic of algorithm OICKM, select one of hot topics of key as K-Means algorithm clustering center; For new news text information using Single-Pass topic classification algorithm, then using similarity formula new add text information and the similarity of original text messages, thus reducing the amount of calculation in the process of clustering, so as to speed up the discovery of the hot topics. Finally, through the experiment and the method of evaluation of news text information on OICKMSP algorithm proposed in this paper and OICKM speed and accuracy of the algorithm are compared.