Comparison of influence metrics in information diffusion networks

The importance of social networks has pushed the analysis of the interactions among their users. The fact that a user may influence their friends or the friends of their friends makes the study of these social interactions necessary for many disciplines. We design a flexible model that can be mapped to many social-like networks. Specifically, we study and compare three of these networks: Enron, Twitter and Scopus. We present a comparison of several variables of influence and how important they are in each network. In addition, we study how correlated they are and the depth of influence propagation. We observe that in general variables exhibit correlation between them but they are not strongly correlated when we study the most influencing users. Finally, the influence is not very deep in all cases which means that influence propagation is generally local.