Design And Implementation Of The Text Orientation Analysis In The Public Opinion Monitoring System

Public opinion reflects the view of the incident and expresses the public’s demands. It is of great significance to keep abreast of the situation of public opinion and lead the public opinion properly. With the development of the Internet, network has become an important way for people to access information and expresses their views. Chairman Hu Jintao says that "The internet has become the collection and distribution center of the ideological and cultural and the amplifier of the public opinion. Public opinion on the internet has the following features:affecting large number of people and areas, happening in sudden, public opinion information always contains incendiary and orientation, public opinion information is always one-sided. Because of the characteristics, it is important to monitor the public opinion on the internet and lead it correctly. Public opinion information is usually of orientation. It will provide valuable reference for analyzing public opinion if we can identify the orientation of the public’s topic. This paper researches the text orientation analysis and proposes a new text orientation calculation algorithm based on WAF(Word Activation Force). The experiments show that this method performs really well and it has been used in the public opinion monitoring system.The main work of this paper is as follows:The first, design processing flow, storage and overall architecture of a public opinion monitoring system which processes mass internet data. Design the system framework, loading balance between several servers, mechanism of data synchronization between servers and database storage scheme for the need of mass data fast processing. The second discusses the design and implementation of some modules in the monitoring system faces mass data including removing duplicate data collected from internet, Chinese words segments, clustring module, hot words module, sensitive words module, training module for classification, orientation identifying, data table design, database interface for all modules. For the mass data, introduces the program of data batch updates of Oracle database and the spliting table storage program of MongoDB database. The third, by studying the shortcomings of traditional text orientation analysis methods, proposes a more effective text orientation recognition method based on WAF. Traditional method usually cannot reflects the overall situation of the orientation of the documentation set, however the method based on WAF can find the orientation characteristics of the documentation set effectively and performs better in recognizing the overall situation of the orientation of the documentation set. Experiments between algorithm based on WAF and algorithm based on orientation weight proves the effectiveness of this method.