GNNPapers

GNNPapers is a curated reading list of academic papers on graph neural networks, abbreviated as GNN. A graph neural network is a type of machine learning model designed to work on data that is naturally shaped like a network — nodes connected by edges — rather than on plain images, text, or tables. Examples include social networks, knowledge graphs, molecules, and road maps.

The repository itself is not a piece of software. It is a single long README that lists must-read papers under thematic sections, with each entry showing the paper title, the conference or journal it appeared in, the year, a link to the paper, and the authors. The contents are organised into three main parts. The first is surveys, which give broad overviews of the field. The second is models, broken down into basic architectures, graph types, pooling methods, analysis, efficiency, and explainability. The third is applications, where graph neural networks have been tried out, including physics, chemistry and biology, knowledge graphs, recommender systems, computer vision, natural language processing, generation, combinatorial optimisation, adversarial attack, graph clustering, reinforcement learning, traffic networks, few-shot learning, program representation, and social networks.

You would visit this repository when you are starting research or self-study in graph neural networks and want a structured entry point, when you want to follow how the field has evolved, or when you are writing a literature review and need a starting bibliography. The list was contributed by researchers Jie Zhou, Ganqu Cui, Zhengyan Zhang, and Yushi Bai. The full README is longer than what was provided.