Rank-Based Models of Network Structure and the Discovery of Content

Abstract

Research on self-organizing networks, especially in the context of the Web graph, holds great promise to understand the complexity that underlies many social systems. We argue that models of social network structure should begin to consider how structure arises from the “content” of networks, a term we use to describe attributes of network actors that are independent of their structural position, such as skill, intelligence, or wealth. We propose a rank model of how content (operationalized as attribute rank relative to other individuals) may change amongst agents over time within a stochastic system. We then propose a model of network self-organization based on this rank model. Finally, we demonstrate how one may make inferences about the content of networks when attributes are unobserved, but network structures are readily measured. This approach holds promise to enhance our study of social interactions within the Web graph and in complex social networks in general.