Measuring What Matters: A Unified Evaluation Framework for GNN Explainability
2026-07-06 • Machine Learning
Machine Learning
AI summaryⓘ
The authors focus on making explanations for Graph Neural Networks (GNNs) clearer and more reliable. They point out that many explanation methods exist, but it’s hard to know which to trust or use. To fix this, they create a new testing framework that doesn’t rely on knowing the ‘right answer’ beforehand and measures explanations based on graph structure and features. Their study finds several top-performing explanation methods but shows no single method works best for everything. Finally, they offer practical advice for people building trustworthy GNN systems.
Graph Neural NetworksExplainable AIGraph explainabilityBenchmarkingTopological structureNode featuresPareto frontTrustworthinessEvaluation metricsMachine Learning pipelines
Authors
Francesco Paolo Nerini, Mirko Zaffaroni, Paolo Baracco, Gabriele Ciravegna, Alan Perotti
Abstract
Graph eXplainable AI (G-XAI) is increasingly important for making Graph Neural Networks interpretable and accountable. While a growing number of explainers are available, choosing the right method and assessing the trustworthiness of its outputs remains unclear. Consistent evaluation practices and actionable guidance are still missing, hindering practical adoption. In this paper, we introduce a unified, quantitative benchmarking framework for G-XAI that requires no ground-truth assumptions. We formalize tabular explainability metrics for graph data, evaluating topological structure and node features as independent components. Our large-scale benchmarking study identifies explainers that consistently lie on the Pareto front across metric pairs and tasks, establishing robustly non-dominated solutions - while confirming that no single explainer achieves universal superiority. We distill our findings into actionable G-XAI usability guidelines to support Machine Learning practitioners in evaluating and deploying trustworthy GNN-based pipelines.