LHRLAB/HyperGraphRAG

[NeurIPS 2025] Official resources of "HyperGraphRAG: Retrieval-Augmented Generation via Hypergraph-Structured Knowledge Representation".

What it solves

HyperGraphRAG addresses the limitations of traditional Retrieval-Augmented Generation (RAG) by using a hypergraph-structured knowledge representation. This allows the system to capture more complex, multi-entity relationships that standard graphs cannot represent effectively, improving the quality of information retrieval for complex queries.

How it works

The system constructs a Knowledge HyperGraph from provided text contexts. It then uses this hypergraph structure to retrieve relevant information when a user submits a query, allowing the LLM to generate answers based on a hypergraph-structured knowledge base.

Who it’s for

Researchers and developers building RAG systems that need to handle complex, high-order relationships between entities in their data.

Highlights

  • Hypergraph-Structured Knowledge: Uses hypergraphs instead of simple graphs to represent knowledge.
  • NeurIPS 2025: Based on a peer-reviewed research paper published at NeurIPS 2025.
  • Simple API: Provides a straightforward insert and query interface for building and querying the knowledge base.