Degree-degree dependency measures in directed networks

Pim van der Hoorn (UT)


Dependencies between the degrees on both sides of a random edge are an important characteristic of network topology. They influence, among others, information spread and social consensus in networks. We show that the measure most commonly used for these dependencies has essential flaws. To remedy this we introduce new, rank based, measures and prove that these are consistent estimators for degree-degree dependencies in random directed networks, under minor conditions on the degree sequences. Finally we investigate these rank correlations in the directed configuration model and show that this model can function as a null model for these measures.