Andrea Lancichinetti (ISI)
Anomalous structures in networks are crucial for understanding the function and the organization of complex systems. The problem is related to computing the probability of finding such structures on a proper null model. Although it is quite natural to choose as a candidate the configuration model in the case of unweighted networks, it is less trivial how to define good null models for weighted and time-dependent networks. We discuss the problem, its connections with studies on backbones of weighted networks, and some implications on how to make null models more similar to true models of real systems.