Design of a Multimodal Intelligent Public Opinion Analysis System

As an emerging research field, multimodal sentiment analysis tasks aim to identify the speaker's emotions by combining information from different modalities. In recent years, it has been increasingly used in areas such as public opinion analysis, intelligent dialogue and user profiling. This article systematically sorts out the development context of multimodal intelligent public opinion analysis technology, and focuses on the application status of minority language processing technology in the field of public opinion monitoring. By analyzing the technical bottlenecks of the current public opinion analysis system, a multimodal fusion solution based on deep learning is proposed, and the effectiveness of the solution in social governance in ethnic minority areas is verified by combining actual cases. The research results show that a multimodal system integrating speech recognition, machine translation and sentiment analysis can significantly improve the efficiency of processing public opinion in minority languages and provide technical support for maintaining social stability in ethnic regions.