Abstract

Online social networks provide a rich substrate for information diffusion. Everyday, a portion of the billions posts and photos shared on Facebook generates cascades of reshares, enabling many additional people to see, like, comment and in turn reshare them. This data provides a unique opportunity to study the way information spreads from one person to another.

This talk is a overview of several research projects that Facebook has been conducting to capture what virality means and imply in terms of social network analysis. From Obama’s hug and expositions of wild rumors, to predicting the size of cascades and campaigns for a million likes in the hopes of getting lucky, we will cover practical aspects of data exploration, sanitization and modeling.

Bio: Adrien Friggeri is a computational sociologist on Facebook’s Data Science team. His research ranges from the abstract study of structural graph features to understanding the links between psychology and network topology. After spending a few years establishing a mathematical model of social communities, his work has shifted back towards dynamic processes and information diffusion on social networks.