This repository is a curated list of academic papers focused on the explainability of Graph Neural Networks (GNNs). It serves as a valuable resource for researchers and practitioners in graph machine learning seeking to understand, evaluate, and develop methods for interpreting GNN models. The collection aims to provide a comprehensive overview of the field, covering surveys, foundational papers, and recent advancements.

How It Works

The repository organizes papers by year and topic, including surveys, specific explanation methods, and benchmark datasets. It highlights key contributions and links to associated publications and code where available. The collection spans various approaches to GNN explainability, from local explanations (e.g., identifying influential nodes or subgraphs) to global explanations (e.g., understanding overall model behavior or learning interpretable concepts).

Quick Start & Requirements

This is a curated list of papers and does not require installation or execution. Links to papers and code repositories are provided within the README.

Highlighted Details

• Comprehensive coverage of GNN explainability research from 2020 to the present.
• Includes links to seminal papers and recent advancements in the field.
• Lists relevant tools and platforms like PyTorch Geometric, DIG, GraphXAI, GraphFramEx, and BAGEL.
• Categorizes papers by explanation type (e.g., surveys, methods, benchmarks) and application domains.

Maintenance & Community

The repository is maintained by flyingdoog. There are no explicit mentions of community channels or active development beyond the curation of papers.

Licensing & Compatibility

The repository itself is not software and does not have a license. The linked papers are subject to their respective publisher licenses.

Limitations & Caveats

This is a static list of papers and does not provide any executable code or tools for GNN explainability. The quality and relevance of the papers are subjective and depend on the user's specific needs.