By collecting video bullet comments to obtain the user’s emotional tendency to the video, it can be used as an emotional label for the video and also provide a reference for subsequent video viewers. In this paper, Python language is used to collect online video bullet comments. After data preprocessing and word segmentation, Baidu AI’s sentence-level sentiment analysis method is called to count and analyze video barrage information. Through subjective and objective emotions, emotional changes along the time axis and other methods, the implicit value or promotion value of video is excavated. This paper takes the popular film and television dramas ‘Dajiang River 3’ and ‘Fanhua’ as examples for empirical analysis. The results show that the current sentiment analysis method can capture the emotional changes with high accuracy in a short time, and reveal that the trend of barrage comments is highly consistent with the evolution of the plot.