Algorithm, Experimentation

The study of collective behavior is to understand how in-dividuals behave in a social network environment. Oceans of data generated by social media like Facebook, Twitter, Flickr and YouTube present opportunities and challenges to studying collective behavior in a large scale. In this work, we aim to learn to predict collective behavior in social me-dia. In particular, given information about some individu-als, how can we infer the behavior of unobserved individ-uals in the same network? A social-dimension based ap-proach is adopted to address the heterogeneity of connec-tions presented in social media. However, the networks in social media are normally of colossal size, involving hun-dreds of thousands or even millions of actors. The scale of networks entails scalable learning of models for collective behavior prediction. To address the scalability issue, we propose an edge-centric clustering scheme to extract sparse social dimensions. With sparse social dimensions, the social-dimension based approach can efficiently handle networks of millions of actors while demonstrating comparable predic-tion performance as other non-scalable methods