Research on Dimension Reduction Method of Public Opinion Short Text Representation Model Based on Combined Neural Network

Abstract

With the rapid development of the Internet and the popularization of social software such as Twitter and Weibo, online public opinion has an increasing influence on public opinion in the entire society. In order to effectively prevent and control vicious incidents, real-time monitoring of online public opinion is becoming increasingly important. important. According to the characteristics of short-sentence information in public opinion, this paper proposes an automatic clustering method based on combinatorial neural network to construct a short-sentence representation model of public opinion, and construct word clusters based on the semantic similarity of characteristic words. The test results show that: in the process of mass text analysis and public opinion discovery, this method can effectively reduce the accuracy of the short text public opinion construction representation model by reducing the dimension of the text representation model, while ensuring the real-time acquisition of public opinion while greatly improving Public opinion finds efficiency.