Asymmetric access to information impacts the power-law exponent in networks

TL;DR

This paper introduces a new network-generation model that accounts for asymmetric information access, revealing its influence on the power-law degree distribution exponent in real-world networks.

Contribution

It presents a novel mechanism incorporating asymmetric information accessibility into network growth, explaining variations in the power-law exponent observed in real networks.

Findings

Asymmetric information access affects the power-law exponent.

The model aligns with empirical data from citation, hyperlink, and social networks.

It explains the diversity of degree distributions in real-world networks.

Abstract

The preferential attachment (PA) process is a popular theory for explaining network power-law degree distributions. In PA, the probability that a new vertex adds an edge to an existing vertex depends on the connectivity of the target vertex. In real-world networks, however, each vertex may have asymmetric accessibility to information. Here we address this issue using a new network-generation mechanism that incorporates asymmetric accessibility to upstream and downstream information. We show that this asymmetric information accessibility directly affects the power-law exponent, producing a broad range of values that are consistent with observations. Our findings shed new light on the possible mechanisms in three important real-world networks: a citation network, a hyperlink network, and an online social network.