The advent of social media is disrupting traditional mechanisms for spreading information. In this "second age of information," the economy of attention drives formidable forces to impact opinions, culture, policy, and advertising profit. Yet the dynamical processes that drive popularity in our online world are still largely unexplored. In the Truthy.indiana.eduproject, we are studying how memes propagate through Twitter with a focus on the detection of misleading political information. This talk outlines a simple framework to capture and model various dynamic views of online traffic and popularity by mining streams of social media events. A first application is the study of popularity dynamics in Wikipedia and the Web at large. The temporal and magnitude behaviors of popularity follow statistical laws typical of critical processes, such as earthquakes and avalanches. We offer a simple model that mimics the exogenous shifts of user attention, recovering the key features observed in the empirical data. Second, we analyze Web traffic networks from large samples of Web users. Since simple Markovian models fail to capture both aggregate and individual traffic patterns, we propose an agent-based model that incorporates essential elements of memory and topicality. Finally, we introduce a model of the competition for attention in social media. A dynamic of information diffusion emerges from this process, where a few ideas go viral while most do not. Surprisingly, one can reproduce the massive heterogeneity in the popularity and persistence of ideas without the need to assume different intrinsic values among those ideas.
Joint work with Alessandro Flammini, Michael Conover, Jacob Ratkiewicz, Bruno Gonçalves, Karissa McKelvey, Lilian Weng, Mark Meiss, Alex Rudnick, Luca Aiello, Przemyslaw Grabowicz, Santo Fortunato, José Ramasco, Clayton Davis, Johan Bollen, and Alessandro Vespignani.
This project is supported by the National Science Foundation, McDonnell Foundation, and DARPA. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of these funding agencies.