Research On Network Public Opinion Monitoring System Based On Sensitive Information Mining

With the continuous development of network and the number of Internet users is increasing, more and more people are willing to express their views on the Internet. The Internet has become the important place for communication and public opinion. Network public opinion plays a more and more important role in social life. Timely, comprehensive grasp of network public opinion information not only can effectively prevent the network sudden events, but it also can provide a reference for the administrative departments to make decision scientifically. Strengthen the management of monitoring network public opinion has become a practical problem, which need to be solved by the current government departments at all levels. It is important to establish an effective network public opinion analysis system for the maintenance of the social stability and development.The traditional network public opinion monitoring system discovers network public opinion passively. When it is in the face of massive network information, not only the timeliness, accuracy is low, and it is difficult to respond to the change of public opinion quickly. According to the characteristics of network public opinion formation, which is rapid, has large amount of information, changes quickly, put forward strategies for effective tracking and monitoring. In the research of network public opinion information monitoring system, we focus on the sensitive information capture, hot spots found and topic tracking.The demand for automatic discovery of public opinion hotspot and in-depth analysis of technology research, proposed the establishment of sensitive information analysis framework based on Data Mining. We designed and implemented the definition of the sensitive information, and the sensitivity of the described method. Sensitive information is composed of a sensitive word, word association, correlation degree and correlation. Using data mining techniques (Association Rules), correlation analysis and cluster analysis, returns associated information with sensitive information to the user. It is the effective solution to network hot spot automatic discovery and hot topic tracking. At the same time, due to the large amount of data of network public opinion, in order to improve the efficiency and accuracy of the system, we adopted the effective extraction of text features webpage cleaning technology. The user can understand what is happening on the internet, and understand the whole event quickly, completely and comprehensively. The method described in this article can significantly improve the public opinion monitoring, enhances the usefulness of the system.