Effect of in/out-degree correlation on influence degree of two contrasting information diffusion models

Abstract

How the information diffuses over a large social network depends on both the model employed to simulate the diffusion and the network structure over which the information diffuses. We analyzed both theoretically and empirically how the two contrasting most fundamental diffusion models, Independent Cascade (IC) and Linear Threshold (LT) behave differently or similarly over different network structures. We devised two rewiring structures, one preserving in/out-degree correlation and the other changing in/out-degree correlation while both preserving their in/out-degree distributions, and analyzed how co-link rate and in/out-degree correlation affect the influence degree of each diffusion model using two real world networks, each as the base network on which rewiring is imposed. The results of the theoretical analysis qualitatively explain the empirical results, and the findings help deepen the understanding of complex diffusion phenomena. © 2012 Springer-Verlag.