awesome-explainable-graph-reasoning

Why we included this project

Graph neural networks are hard to trust when you cannot tell why they made a call, and this list is a good place to start if that is your problem. It brings together the research literature and the code, split into sections on explainable predictions, explainable reasoning, theory and survey papers, and a software chapter. The organization lets you move from a survey to a working implementation when you need to audit a model or add interpretability to a pipeline. AstraZeneca maintains it, and it is especially handy for teams comparing GNN explainer methods in areas like drug discovery or molecular property prediction, where knowing why a model decided something matters as much as getting the answer right. It is a reading guide rather than a tool you deploy, so use it to map the field before picking a library.