Analyzing the Characteristics of Ego-Network Structures of Influential Twitter Users

We investigate the structural characteristics of ego networks of influential Twitter users. While existing studies have elucidated the global characteristics of influencers in a social network, it remains unclear what differences exist in the local structure of social networks between influencers and non-influencers. Using four Twitter datasets, we calculate two types of influence scores for each user, called direct influence and indirect influence. Then, we compare several measures of ego networks between users with high influence scores (influencers) and everyone else (non-influencers). Our results show that influencers with strong direct influence have relatively lower values for their average degree, clustering coefficient, and reciprocity, and higher centralization than non-influencers, which suggests that influencers have different ego-network structures than non-influencers. In contrast, our results also show that ego-network structures of users with strong indirect influence are not very different from those of other users, which suggests that the ego-network structure of a user is not useful for estimating their indirect influence.