DEEP-PolyU/Awesome-GraphRAG

Awesome-GraphRAG: A curated list of resources (surveys, papers, benchmarks, and opensource projects) on graph-based retrieval-augmented generation.

What it solves

Traditional retrieval-augmented generation (RAG) retrieves text chunks by similarity, which struggles with multi-hop reasoning and capturing relationships between entities. GraphRAG addresses this by using graph structures to represent knowledge, enabling more accurate and context-aware responses. This repository provides a structured overview of GraphRAG research, helping users quickly find relevant work.

How it works

The repository is an "awesome" list, categorizing research papers into three main areas: knowledge organization (how graphs are built and structured), knowledge retrieval (how relevant information is found on the graph), and knowledge integration (how the retrieved knowledge is fused into the LLM). It also includes a trend chart, benchmarks, and open-source projects. The classification follows the authors' survey paper, and the list is continuously updated.

Who it's for

Researchers, developers, and practitioners working on large language models, information retrieval, and knowledge graphs. It serves as a starting point for understanding GraphRAG and finding relevant resources for building or improving GraphRAG systems.

Highlights

  • Organized into three core areas: knowledge organization, retrieval, and integration.
  • Includes a trend chart showing the development of GraphRAG research.
  • Provides sections for benchmarks and open-source projects.
  • Continuously updated with recent papers and news (e.g., acceptances at top conferences).

Related

  • Project
  • Project
  • Project
  • Project