Multimodal Emotion Analysis of Online Product Marketing Information in Social Interaction and Internet Behavior Based on Generative Artificial Intelligence

With the proliferation of multimodal data such as text, images and videos on social networks, it is particularly important to extract valuable information from these diverse data sources. In this study, a multimodal sentiment analysis framework based on generative artificial intelligence (AI) technology is proposed, aiming at automatically identifying the emotional tendency in online product marketing information. By combining the latest generative AI technology, the model can not only analyze the direct feedback from users, but also capture the potential emotional color. The experimental results show that compared with the traditional single-modal sentiment analysis method, the accuracy of the proposed multimodal model is improved by 15%, and the recall rate is improved by 12%. This shows that the model can provide more accurate market insight for enterprises, and help to better understand customer needs, improve product design and optimize marketing strategies. The demonstrated multimodal sentiment analysis framework provides a powerful tool for understanding and responding to consumer sentiment. Future research will further explore how to enhance the generalization ability of the model and how to deal with emotional differences in cross-cultural background more effectively to achieve a wider application in the global scope.