On the Impact of Communities on Semi-supervised Classification Using Graph Neural Networks

TL;DR

This paper investigates how community structures in graphs influence the effectiveness of Graph Neural Networks in semi-supervised node classification, revealing that community effects vary based on label distribution and providing guidelines for model selection.

Contribution

It systematically analyzes the impact of community structures on GNN performance and offers insights and guidelines for model selection based on graph properties.

Findings

Communities significantly affect GNN classification performance.

Breaking community structure causes performance drops when labels align with communities.

When labels are independent of communities, graph structure has minimal impact.

Abstract

Graph Neural Networks (GNNs) are effective in many applications. Still, there is a limited understanding of the effect of common graph structures on the learning process of GNNs. In this work, we systematically study the impact of community structure on the performance of GNNs in semi-supervised node classification on graphs. Following an ablation study on six datasets, we measure the performance of GNNs on the original graphs, and the change in performance in the presence and the absence of community structure. Our results suggest that communities typically have a major impact on the learning process and classification performance. For example, in cases where the majority of nodes from one community share a single classification label, breaking up community structure results in a significant performance drop. On the other hand, for cases where labels show low correlation with communities, we find that the graph structure is rather irrelevant to the learning process, and a feature-only baseline becomes hard to beat. With our work, we provide deeper insights in the abilities and limitations of GNNs, including a set of general guidelines for model selection based on the graph structure.