HybridRAG-Bench Documentation
HybridRAG-Bench is a benchmarking framework for retrieval-intensive, multi-hop reasoning over hybrid knowledge. It evaluates how retrieval-augmented methods use both unstructured text and structured knowledge graphs under realistic settings.
The framework supports:
Time-framed corpus collection from scientific sources.
Hybrid knowledge construction from aligned text chunks and KG structure.
Reasoning-grounded QA generation across multiple reasoning types.
Automated quality control and reproducible benchmark evaluation.
Documentation