GraphFramEx: Towards Systematic Evaluation of Explainability Methods for Graph Neural Networks
This work addresses the need for a common evaluation protocol for GNN explainability methods, particularly for users in mission-critical applications, though it is incremental as it builds on existing explainability techniques.
The paper tackles the lack of systematic evaluation for explainability methods in Graph Neural Networks (GNNs) by proposing GraphFramEx, a framework that introduces a unique metric combining fidelity measures and classifies explanations based on sufficiency and necessity, finding that shallow techniques like personalized PageRank perform best on synthetic benchmarks, while gradient-based methods excel on complex graphs with meaningful features, but no method dominates across all dimensions.
As one of the most popular machine learning models today, graph neural networks (GNNs) have attracted intense interest recently, and so does their explainability. Users are increasingly interested in a better understanding of GNN models and their outcomes. Unfortunately, today's evaluation frameworks for GNN explainability often rely on few inadequate synthetic datasets, leading to conclusions of limited scope due to a lack of complexity in the problem instances. As GNN models are deployed to more mission-critical applications, we are in dire need for a common evaluation protocol of explainability methods of GNNs. In this paper, we propose, to our best knowledge, the first systematic evaluation framework for GNN explainability, considering explainability on three different "user needs". We propose a unique metric that combines the fidelity measures and classifies explanations based on their quality of being sufficient or necessary. We scope ourselves to node classification tasks and compare the most representative techniques in the field of input-level explainability for GNNs. For the inadequate but widely used synthetic benchmarks, surprisingly shallow techniques such as personalized PageRank have the best performance for a minimum computation time. But when the graph structure is more complex and nodes have meaningful features, gradient-based methods are the best according to our evaluation criteria. However, none dominates the others on all evaluation dimensions and there is always a trade-off. We further apply our evaluation protocol in a case study for frauds explanation on eBay transaction graphs to reflect the production environment.