Multi-Modal Emotion Recognition for Online Education Using Emoji Prompts

Abstract

Online education review data have strong statistical and predictive power but lack efficient and accurate analysis methods. In this paper, we propose a multi-modal emotion analysis method to analyze the online education of college students based on educational data. Specifically, we design a multi-modal emotion analysis method that combines text and emoji data, using pre-training emotional prompt learning to enhance the sentiment polarity. We also analyze whether this fusion model reflects the true emotional polarity. The conducted experiments show that our multi-modal emotion analysis method achieves good performance on several datasets, and multi-modal emotional prompt methods can more accurately reflect emotional expressions in online education data.