Understanding GNNs and Homophily in Dynamic Node Classification

2025-04-29

AI Summary

This work addresses the poorly understood impact of homophily on dynamic node classification performance in graph neural networks (GNNs) applied to dynamic graphs. We introduce—*for the first time*—the notion of *dynamic homophily*, defined as the probability that a node’s future label matches the labels of its current neighbors. We theoretically model and empirically analyze the discriminative capacity of GCN-style models under this measure. Extensive evaluation across multiple dynamic graph benchmarks reveals that mainstream GNNs exhibit strong dependence on dynamic homophily: their performance degrades sharply when this measure is low. Our study bridges a critical gap in extending classical (static) homophily theory to temporal graphs, providing both an interpretable theoretical foundation and practical design principles for developing, evaluating, and enhancing the robustness of dynamic GNNs.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Reasoning under Uncertainty: Graphical Models

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Social Networks and Social Media: Influence propagation, information diffusion, and the prediction on networks
• Web Mining and Content Analysis: Models for Web evolution

Abstract

Homophily, as a measure, has been critical to increasing our understanding of graph neural networks (GNNs). However, to date this measure has only been analyzed in the context of static graphs. In our work, we explore homophily in dynamic settings. Focusing on graph convolutional networks (GCNs), we demonstrate theoretically that in dynamic settings, current GCN discriminative performance is characterized by the probability that a node's future label is the same as its neighbors' current labels. Based on this insight, we propose dynamic homophily, a new measure of homophily that applies in the dynamic setting. This new measure correlates with GNN discriminative performance and sheds light on how to potentially design more powerful GNNs for dynamic graphs. Leveraging a variety of dynamic node classification datasets, we demonstrate that popular GNNs are not robust to low dynamic homophily. Going forward, our work represents an important step towards understanding homophily and GNN performance in dynamic node classification.

Problem

• Analyzing homophily in dynamic graphs for GNNs
• Proposing dynamic homophily to measure GNN performance
• Improving GNN robustness in dynamic node classification

Innovation

• Proposes dynamic homophily for dynamic graphs
• Analyzes GCN performance in dynamic settings
• Demonstrates GNN robustness to dynamic homophily