Empirical Analysis and Evolving Model of Bipartite Networks

TL;DR

This paper analyzes real-world bipartite networks, classifies them into dependence and independence types, and proposes models to explain the scale-free properties observed in dependence networks.

Contribution

It introduces a new classification of bipartite networks and proposes models that explain the scale-free degree distribution without relying on preferential attachment.

Findings

Actors in dependence networks exhibit scale-free degree distribution.

Independence networks do not show a consistent degree distribution.

Proposed models qualitatively match empirical observations.

Abstract

Many real-world networks display a natural bipartite structure. Investigating it based on the original structure is helpful to get deep understanding about the networks. In this paper, some real-world bipartite networks are collected and divided into two types, dependence bipartite networks and independence bipartite networks, according to the different relation of two sets of nodes. By analyzing them, the results show that the actors nodes have scale-free property in the dependence networks, and there is no accordant degree distribution in the independence networks for both two types of nodes. In order to understand the scale-free property of actors in dependence networks, two growing bipartite models without the preferential attachment principle are proposed. The models show the scale-free phenomena in actors' degree distribution. It also gives well qualitatively consistent behavior with the empirical results.