Understanding the Role of Video Quality and Emotion in Live Streaming Viewership

Live streaming platforms, such as Twitch and Youtube Live, have become the most prevalent user-generated content platforms globally. Despite the prevalence, scholars and practitioners have spent much effort exploring the engagement pattern and the culture of live streaming communities. In this paper, we investigate the effect of video quality and streamers’ emotional expression on real-time viewership. We first propose a context-aware deep learning model to identify the camera frame and extract video quality and emotion traits from live streaming videos with state-of-art computer vision implementations. We then empirically examine the correlation between the proposed measures and real-time viewership. The preliminary result suggests that live streaming users generally prefer live streaming content with more pleasant valence, less thrilling elements, and “warm” colors. This research takes the first step to examine live streaming videos with the sectional settings and significantly contributes to the live streaming literature from both theoretical and technical perspectives.