Awesome-GraphRAG
A curated list of surveys, papers, benchmarks, and open-source projects on graph-based retrieval-augmented generation, organized around a published GraphRAG survey.
Awesome-GraphRAG collects resources on GraphRAG, retrieval-augmented generation that incorporates graph-structured knowledge, organized according to the categories used in an accompanying academic survey paper. It distinguishes non-graph RAG, which ranks and retrieves relevant text chunks by similarity, from knowledge-based GraphRAG, which extracts entity-relation knowledge graphs for fine-grained retrieval, and index-based GraphRAG, which summarizes a corpus into a linked index of topic nodes.
The list is continuously updated with newly accepted papers from the maintainers' own research group, spanning topics such as memory-enhanced RAG, RAG faithfulness, reliable legal reasoning, adversarial attacks on GraphRAG, efficient graph construction, and benchmark datasets for evaluating GraphRAG systems, alongside the broader published survey literature. It is MIT licensed.