Graph RAG

Graph RAG (Knowledge Graph Retrieval Augmented Generation) is a technique for richly understanding text datasets by combining text extraction, network analysis, and LLM prompting and summarization into a single end-to-end system. Microsoft Research's approach creates a knowledge graph based on an input corpus, where this graph, along with community summaries and graph machine learning outputs, are used to augment prompts at query time. GraphRAG uses a large language model (LLM) to automate the extraction of a rich knowledge graph from any collection of text documents and features the ability to report on the semantic structure of the data prior to any user queries. This methodology uses knowledge graph memory structures to enhance LLM outputs, enabling sophisticated reasoning over complex, interconnected information that traditional RAG systems struggle to process effectively.

Related terms

• Knowledge Graph RAG
• LangChain Graph RAG
• Agentic Retrieval-Augmented Generation: a survey on agentic RAG
• AI what is RAG
• C-RAG: Certified Generation Risks for Retrieval-Augmented Language Models

Vstorm builds production systems that use Graph RAG: RAG development service, Agentic AI consulting.

← Back to the AI Glossary