Who watches (and shares) what on youtube? and when?: using twitter to understand youtube viewership

We combine user-centric data from Twitter with video-centric data from YouTube to analyze who watches and shares what on YouTube. The combination of these data sets1, with around 87k Twitter users, 5.6 million YouTube videos and 15 million video sharing events, allows rich analysis that goes beyond what could be obtained with either of the two data sets individually. For Twitter, we generate user features that relate to (i) activity, (ii) interests, and (iii) demographics. For YouTube, we obtain video features for (i) the topic, (ii) the popularity, and (iii) the polariza-tion. These two feature sets are then combined through sharing events for YouTube URLs on Twitter. This combination is done both in a user-centric, a video-centric and a sharing-event-centric manner. For the user-centric analysis, we show how Twitter user features correlate both with YouTube features and with sharing-related fea-tures. As two examples, we show that urban users are quicker to share than rural users and that for some notions of “influence”, but not others, influential users on Twitter share videos with a higher number of views. For the video-centric analysis, we find a superlinear relation be-tween initial twitter shares, and the final amounts of views, showing the correlated behavior of Twitter. Regarding user impact, we find that the total amount of followers of the users that shared the video in the first week does not play a role on its final popularity. On the other hand, aggregated user retweet rates serve as a better predictor for video popularity in YouTube. For the sharing-centric analysis, we reveal the existence of cor-related behavior concerning the time between video creation and sharing within certain timescales, showing the time onset for a co-herent response, and the time limit after which collective responses are extremely unlikely. We show that response times depend on the category of the video, revealing that Twitter sharing of a video is highly dependent on its content. 1The dataset is available at