Multi-Dimensional Visual Analysis of Digital Media Emotion Based on Machine Learning

In the context of the rapid growth of social media and short video platforms, how to accurately capture public sentiment from massive graphic videos has become a key problem. Existing tools often only analyze a single type of data, and the results are static and lack of interaction, which makes it difficult to support deep insight. Therefore, the authors have developed a new emotional analysis framework, which can “read” words, “understand” images, and “perceive” time changes at the same time, and fuse this information through intelligent algorithms to reduce the understanding deviation between modes. The system is equipped with an intuitive visual interface, which supports users to zoom, filter, and track the trajectory of emotional changes like a map. The test results show that this method is superior to the traditional scheme in accuracy and user experience. This technology can help enterprises to dynamically monitor brand reputation and use generative AI to create virtual content with emotional tension to thereby supporting more advanced and intelligent interactive experiences.