Research on Identification of Network Public Opinion Information based on Graph Convolutional Networks

Abstract

Traditional public opinion information identification methods have poor performance, eitherlow accuracy, or rely on hand-designed features. This paper converts public opinion information identification to text classification problem, and proposes a public opinion information identification method based on Word2Vec and graph convolutional networks. First, Word2Vec is used to train word vector and word-article graphs are constructed; then, the graphs are trained and classified by graph convolutional neural network; finally, network public opinion information recognition is completed according to the classification results. The experimental results on the constructed Central Asian country data set show that the proposed method has achieved better performance,where the average identification accuracy of “Belt and Road” network public opinion information reached 85.58%.Furthermore, the performance on other data sets is also comparable to current mainstream text classification methods.