The diffusion of political memes in social media: keynote abstract

This talk presents ongoing work on the study of information diffusion in social media, focusing in particular on political communication in the Twitter microblogging network. Social media platforms play an important role in shaping political discourse in the US and around the world. The truthy.indiana.edu infrastructure allows us to mine and visualize a large stream of social media data related to political themes. The analyses in this keynote address polarization and cross-ideological communication, and partisan asymmetries in the online political activities of social media users. Machine learning efforts can successfully 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. 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. Related to political communication are questions about the birth of online social movements. Social media data provides an opportunity to look for signatures that capture these seminal events. 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.