Query-Aware Flow Diffusion for Graph-Based RAG with Retrieval Guarantees
The paper introduces Query-Aware Flow Diffusion for Graph-Based Retrieval-Augmented Generation (QAFD-RAG), a new framework designed to enhance graph traversal based on query semantics. This approach aims to improve the relevance of retrieved subgraphs while providing statistical guarantees for their quality. Experimental results indicate that QAFD-RAG consistently outperforms existing graph-based methods in tasks such as question answering and text-to-SQL.
- ▪QAFD-RAG dynamically adapts graph traversal to align with the holistic semantics of each query.
- ▪The framework provides statistical guarantees for the quality of retrieved subgraphs under mild conditions.
- ▪Experiments show that QAFD-RAG achieves significant improvements over state-of-the-art graph-based RAG methods.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Information Retrieval arXiv:2605.18775 (cs) [Submitted on 21 Apr 2026] Title:Query-Aware Flow Diffusion for Graph-Based RAG with Retrieval Guarantees Authors:Zhuoping Zhou, Davoud Ataee Tarzanagh, Sima Didari, Wenjun Hu, Baruch Gutow, Oxana Verkholyak, Masoud Faraki, Heng Hao, Hankyu Moon, Seungjai Min View a PDF of the paper titled Query-Aware Flow Diffusion for Graph-Based RAG with Retrieval Guarantees, by Zhuoping Zhou and 9 other authors View PDF HTML (experimental) Abstract:Graph-based Retrieval-Augmented Generation (RAG) systems leverage interconnected knowledge structures to capture complex relationships that flat retrieval struggles with, enabling multi-hop reasoning.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
