This paper presents a new, socially-informed approach for characterizing the connectivity structure among Twitter users. We primarily focus on a sub-graph of top 10K most-followed users (or elites) that we refer to as elite network. We present a new technique for efficiently capturing the Twitter elite network along with social attributes of individual elite nodes. We show that the elite network (even at smaller sizes) is composed of a 15-20 resilient elite communities that all exhibit a clear social cohesion. These characteristics imply that the elite communities represent "socially meaningful" components of the Twitter structure and offer a coarse view of the Twitter elite network.

We then characterize the community-level structure of the elite network and identify the pairwise tendencies between elite communities to follow each other. We also assess the cross-influence between elite communities based on retweeting and replying and show that such influences are effectively contained within individual elite communities. Finally, we illustrate that most regular (non- elite) Twitter users tend to primarily follow (i.e., show interest to) users in a single elite community. A group of regular users who primarily follow an elite community form its "shadow partition". We show that the fraction of relationships between elites that span across elite communities is very similar to the fraction of relationship between regular users that span across different shadow partitions. This suggests that elite communities sketch a socially-aware and coarse view of the entire Twitter structure.