An Analysis on The Network Structure of Influential Communities in Twitter

Abstract

Over the past years online social networks have become a major target for marketing strategies, generating a need for methods to efficiently spread information through these networks. Close knit communities have developed on these platforms through groups of users connecting with like minded individuals. In this thesis we use data pulled from Twitter's API and from simulations designed to mirror the Twitter network to pursue an in depth analysis of the network structure and influence of these communities. Through this analysis we draw several conclusions. First, the influence of users in these communities is correlated to the total number of followers in their neighborhood. Second, influential communities tend to be more tightly clustered than other areas of the network. Using these observations, we develop an algorithm to detect influential communities in Twitter and show that correctly prioritizing connections yields significant gains in message visibility.