Graph RAG
An advanced retrieval technique that enhances large language models by structuring knowledge into interconnected entities and relations instead of flat texts.
Think of It Like This
Like studying for a history test by drawing a complex mind map of how historical figures influenced each other, rather than just reading paragraphs in order.
Graph Retrieval-Augmented Generation extracts entities and their relationships from source documents to construct a knowledge graph. During inference, it traverses this graph to retrieve highly contextual and synthesized information. This approach excels at answering global, multi-hop questions that traditional vector-based semantic search often struggles to resolve.