PCA-SVM Monitoring Model And Application Of Network Public Opinion

With the violent development of Internet information technology,the scale of Internet user base continues to expand.Various forms of social forum applications are more common.The Internet has become an important carrier for disseminating information.Due to the free,open,convenient,instant and hidden characteristics of the Internet,netizens will fully express their opinions and obtain information in a timely manner on social forums.Some social hotspot issues or events containing potential crises will spread rapidly in this way.A large number of comments and opinions with emotional tendencies published by netizens will gather together to form network public opinion.Monitoring online public opinion will help prevent the spread of bad information and guide the wrong and untrue public opinion correctly.It is of great significance for understanding social conditions and public opinion,creating a healthy Internet environment and leading the public to create a good social atmosphere.This paper describes the research background and significance of the network public opinion,summarizes the results at home and abroad,and combs the theoretical knowledge and analysis methods of the network public opinion monitoring.It fully considers the development characteristics and evolution process of network public opinion,constructs a network public opinion monitoring indicator system based on three dimensions: public opinion source,dissemination and audience.On this basis,the principal component factors are extracted,the correlation between the indicators is eliminated,and the purpose of simplifying the research process is achieved.Then,the comprehensive variable is used as the input vector of multi-category SVM.It adopts a one-to-many classification strategy to build a PCA-SVM monitoring model of network public opinion.The empirical analysis of recent representative high-profile online public opinion events verifies the excellent properties of the model.The research results of this paper are as follows: a feasible and effective network public opinion monitoring index system has been established,which enriches the research content of the public opinion index system and has reference significance.It builds a network public opinion monitoring model based on support vector machine and the principal component analysis,at the same time,verifies its effectiveness and accuracy through empirical analysis.It enriches the theoretical framework for dealing with online public opinion events,and provides reference ideas for the public opinion monitoring work of relevant supervision departments.