graphrag
View on GitHubA modular graph-based Retrieval-Augmented Generation (RAG) system
Microsoft GraphRAG is a modular Python pipeline that uses LLMs to extract entities and relationships from unstructured text, build a knowledge graph, cluster communities, and answer queries with graph-based retrieval. Note: the repo is largely in maintenance mode with only bug fixes and CVE updates.
Use Cases
Question answering over private narrative datasetsBuilding knowledge graphs from unstructured documentsGlobal summarization and theming across large corporaEntity and relationship extraction with LLMsCommunity detection and hierarchical summarizationRetrieval-augmented generation with graph contextResearch on graph memory structures for LLM reasoning
Built With
- Language
- Python
- Frameworks
- Microsoft GraphRAG · Python · Pandas · NetworkX · Parquet
Tags
graphrag · knowledge-graph · rag · graph-based-rag · entity-extraction · community-detection · question-answering · llm-pipeline · context-augmentation · unstructured-text · microsoft-research · local-search · global-search