Research on Emotional Expression and Interactive Characteristics of Social Media Based on TF-IDF and Naive Bayes

Abstract

Social media has become the core carrier of human emotional expression and information interaction. There are significant differences in emotional distribution and user interaction characteristics between different platforms. Taking 709 social media texts as the research sample, this paper uses TF-IDF feature extraction and naive Bayes algorithm to build an emotion classification model, combined with descriptive statistics, visual analysis and other methods, to explore the relationship between social media emotional expression differences and emotional types and user interaction behavior. The results show that: social media texts are mainly neutral emotions, and the emotional distribution shows obvious platform specificity; The likes generally prefer positive emotional content, and the platforms of forwarding behavior differ significantly; The accuracy of TF-IDF+naive Bayesian model is 84%, which can effectively realize the automatic emotion recognition of social media text. This study enriches the method system of social media emotion analysis, and provides accurate decision-making reference for content creation, platform operation and brand marketing.