An Analysis of the Streamer Behaviors in Social Live Video Streaming

Social live video streaming systems, for example Facebook Live, Youtube Live, and Twitch.tv, are currently very popular both in industry and in academia. In this article, we provide a detailed analysis of broadcasting patterns in social live streaming based on our publicly available (anonymized) dataset that contains behavioral information on over 5,819 streamers. We found that streamers have distinctive streaming patterns, which we summarize into 5 groups, including professional streamers, regular streamers, casual streamers, hit-and-run streamers and lost professional streamers. We investigate the behavioral differences among them, namely the amount of donations streamers received and the number of danmaku messages streamers received. To alleviate the influence of unrealistic online situation, for our measurement, we analyze the factors that potentially influence the amount of donations streamers received except professional streamers. Among other results, we find that the number of person who have donated free gifts, is positively correlated with the amount of donations the streamers received for regular streamers. In contrast, for other three types of streamers, we conject that there are much audience who maybe prefer to donate free gifts rather than valuable gifts. Our findings and discussions are insightful for improving social live video streaming services.