Research on Public Opinion Analysis Based on Natural Language Processing

This dissertation zeroes in on the research of public opinion analysis grounded in natural language processing. In the research process, text data from multiple channels such as social media and forums are initially gathered as samples for public opinion analysis. Methodologically, the word embedding algorithm is employed to convert texts into vector representations, enabling computers to comprehend text semantics. A deep-learning model integrating Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN) is utilized to conduct sentiment classification on texts. The convolutional layer extracts text features, while the recurrent layer processes text sequence information. The results demonstrate that this model has achieved a relatively high accuracy rate in the task of public opinion sentiment analysis and can effectively distinguish positive, negative, and neutral public opinion inclinations. The research indicates that deeplearning algorithms based on natural language processing possess remarkable advantages in public opinion analysis and can provide robust data support for relevant decision-making.