Research On The Application Of Text Orientation Analysis In The Public Opinion Monitoring System

Public opinion means the political attitudes of majority people towards the state managers about the occurrence and development of some social issues under some kind of social circumstances. With the popularity of the Internet, network public opinion take "multi-media" as the biggest feature, network information abundance of public opinion show "explosive" posture .First, the amount of network public opinion information is huge, and the second is its numerous categories; third, the background information is complicate, especially sudden events and social epidemic events, often cause common concern of various social groups and political forces immediately. In such circumstances, specify and classify all the views by hand is unrealistic. Only by using computer technology to collect and analysis public opinion corpus automatically can establish a comprehensive, effective and fast public opinion monitoring early-warning mechanism.In the traditional public opinion monitoring system, the topic was detected by text clustering method, designers often take little concern about emotional factors of the comment, often only reach the level of hot topic discovery. In recent years, although there is some study on the characteristics of public opinion conducted in-depth analysis, and they presented the definition of some key points of public opinion, but they did not give a specific mathematical model, found no viable options discovery methods more than manual methods. This paper attempts to bring the natural language understanding technology, orientation analysis of the text, into the public opinion monitoring system, studied on the application of the text orientation analysis technology in public opinion monitoring system. This paper first introduced the system design of public opinion monitoring system that with text orientation analysis technology, and then gave the detailed design of the main modules. On the basis of the foundation of word polarity dictionary, this paper studied the feathers of several comment orientation analysis schemes. At last, this paper detect the hot topic by the scale of the comments of netizens participated in the cluster, and presented the detect model of polar topic, focus topic, sensitive topic to enhance the practical value of the public opinion monitoring system.