Tracking the Diffusion of Ideas in Social Media

In this talk I will present ongoing work on the study of information diffusion in social media, focusing in particular on the Twitter microblogging network.  One domain of particular interest is that of politics. Social media platforms play an important role in shaping political discourse in the US and around the world. Our infrastructure allows us to mine a large stream of social media data related to political themes. Our analysis addresses polarization and cross-ideological communication, and partisan asymmetries in the online political activities of social media users. I will discuss successful machine learning efforts that leverage the structure of meme diffusion networks to detect orchestrated astroturf attacks that simulate grassroots campaigns, and to predict the political affiliation of active users. I will show that the retweet network segregates individuals into two distinct, homogenous communities of left- and right-leaning users.  The mention network does not exhibit this kind of segregation, instead forming a communication bridge across which information flows between these two partisan communities.  We propose a mechanism of action to explain these divergent topologies and provide statistical evidence in support of this hypothesis. Finally, I will 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. I will show that the relative popularity of different topics, the diversity of information to which we are exposed, and the fading of our collective interests for specific memes, can all be explained as deriving from a combination between the competition for limited attention and the structure of social networks. 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, Matthew Francisco, Lilian Weng, Mark Meiss, Snehal Patil, Luca Aiello, Przemyslaw Grabowicz, and Alessandro Vespignani.

This project is supported by the National Science Foundation under award CCF-1101743: Meme Diffusion Through Mass Social Media. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.