Networked Inequality: Preferential Attachment Bias in Graph Neural Network Link Prediction

TL;DR

This paper reveals how GCNs exhibit within-group preferential attachment bias in link prediction, leading to fairness issues, and proposes a metric and training strategy to mitigate this bias in real-world networks.

Contribution

The study uncovers the preferential attachment bias in GCN link prediction, introduces a fairness metric, and proposes a training strategy to reduce within-group unfairness.

Findings

GCNs with symmetric normalized filters have within-group preferential attachment bias.

The proposed fairness metric quantifies disparities in link prediction within groups.

The training strategy effectively reduces within-group unfairness in multiple network types.

Abstract

Graph neural network (GNN) link prediction is increasingly deployed in citation, collaboration, and online social networks to recommend academic literature, collaborators, and friends. While prior research has investigated the dyadic fairness of GNN link prediction, the within-group (e.g., queer women) fairness and "rich get richer" dynamics of link prediction remain underexplored. However, these aspects have significant consequences for degree and power imbalances in networks. In this paper, we shed light on how degree bias in networks affects Graph Convolutional Network (GCN) link prediction. In particular, we theoretically uncover that GCNs with a symmetric normalized graph filter have a within-group preferential attachment bias. We validate our theoretical analysis on real-world citation, collaboration, and online social networks. We further bridge GCN's preferential attachment bias with unfairness in link prediction and propose a new within-group fairness metric. This metric quantifies disparities in link prediction scores within social groups, towards combating the amplification of degree and power disparities. Finally, we propose a simple training-time strategy to alleviate within-group unfairness, and we show that it is effective on citation, social, and credit networks.