Multi-modal emotion recognition in e-commerce live broadcast scenarios

Multimodal sentiment analysis is an important branch of natural language processing. This paper reviews multimodal sentiment recognition in live broadcast scenarios. Firstly, the research basis of sentiment analysis and multimodal fusion is reviewed. Then, the research basis is summarized from the three single-modal sentiment analyses of text, expression, and speech. Finally, based on the existing research of multi-modal sentiment analysis and combined with the e-commerce live broadcast scene, the improvement and fusion strategies of three single-modal sentiment analyses in live broadcast are discussed.