Back to Publications
journal · 2025

A Survey of Explainability Methods for Graph Neural Networks

Tunde Okafor, Abhiraj Bibhar · ACM Computing Surveys

Explainable AIGraph Learning

Abstract

Graph neural networks (GNNs) are increasingly deployed in high-stakes settings, yet the tools available for explaining their predictions lag behind those for standard deep networks. This survey organizes the growing landscape of GNN explainability methods along three axes — instance-level versus model-level, perturbation-based versus gradient-based, and faithfulness versus plausibility — and evaluates each family against a consistent set of faithfulness benchmarks.

We find that many popular attribution methods produce explanations that are locally plausible but fail faithfulness tests under systematic perturbation, and we outline a research agenda for methods that are explainable by construction rather than by post-hoc approximation.

Key contributions

  • A unifying taxonomy spanning over 40 GNN explainability methods.
  • A faithfulness benchmark suite applicable across explanation families.
  • An analysis showing common attribution methods often fail faithfulness checks despite appearing plausible.
PDF DOI Publisher
Back to all publications

On this page

Authors

  • T Tunde Okafor
  • A Abhiraj Bibhar (this author)

Cite this paper

Okafor, T. & Bibhar, A. (2025). A Survey of Explainability Methods for Graph Neural Networks. ACM Computing Surveys.

@article{okafor2025a,
  title     = {A Survey of Explainability Methods for Graph Neural Networks},
  author    = {Okafor, Tunde and Bibhar, Abhiraj},
  journal   = {ACM Computing Surveys},
  year      = {2025}
}