A multi-resolution approach to learning with overlapping communities

The recent few years have witnessed a rapid surge of participatory web and social media, enabling a new laboratory for studying human relations and collective behavior on an unprecedented scale. In this work, we attempt to harness the predictive power of social connections to determine the preferences or behaviors of individuals such as whether a user supports a certain political view, whether one likes one product, whether he/she would like to vote for a presidential candidate, etc. Since an actor is likely to participate in multiple different communities with each regulating the actor's behavior in varying degrees, and a natural hierarchy might exist between these communities, we propose to zoom into a network at multiple different resolutions and determine which communities are informative of a targeted behavior. We develop an efficient algorithm to extract a hierarchy of overlapping communities. Empirical results on several largescale social media networks demonstrate the superiority of our proposed approach over existing ones without considering the multi-resolution or overlapping property, indicating its highly promising potential in real-world applications.