Vu Lam Anh

MaGiX: A Multi-Granular Adaptive Graph Intelligence Framework for Enhancing Cross-Lingual RAG

Retrieval-Augmented Generation (RAG) enhances large language models by grounding their outputs in external knowledge. Recent advances in Graph-based RAG (GRAG) frameworks, such as GraphRAG, LightRAG, and HippoRAG2, integrate knowledge graphs into the retrieval process to improve multi-hop reasoning