Research map / Graph learning / Knowledge and temporal graphs
Query-Aware Flow Diffusion for Graph-Based RAG with Retrieval Guarantees
Zhuoping Zhou, Davoud Ataee Tarzanagh, Sima Didari, Wenjun Hu, Baruch Gutow, Oxana Verkholyak, Masoud Faraki, Heng Hao, Hankyu Moon, Seungjai Min
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use query-aware flow diffusion retrieval instead of GraphRAG | QAFD-RAG · UltraDomain (5 shown domains) | Comprehensiveness: Agriculture 87.30 -> 89.93; Legal 84.96 -> 86.19 | Improved |
| QAFD-RAG vs GraphRAG on Legal diversity | QAFD-RAG · UltraDomain Legal | Diversity: 78.74 -> 77.14 | Hurt |
| QAFD-RAG vs best baseline on SQuALITY | QAFD-RAG · SQuALITY | METEOR / BLEU-1: METEOR 24.38 -> 25.59; BLEU-1 34.17 -> 35.44 | Improved |
| QAFD-RAG vs LightRAG on ROUGE-1 | QAFD-RAG · SQuALITY | ROUGE-1 F1: 28.59 -> 28.43 | No clear effect |
| QAFD-RAG vs HippoRAG on HotpotQA | QAFD-RAG · HotpotQA | F1: 58.67 -> 73.42 | Improved |
| QAFD-RAG vs HippoRAG on 2WikiMultihopQA | QAFD-RAG · 2WikiMultihopQA | F1: 70.33 -> 69.41 | No clear effect |
| QAFD-RAG vs HippoRAG on MuSiQue | QAFD-RAG · MuSiQue | F1: 38.33 -> 47.99 | Improved |
| use QAFD-RAG schema linking instead of ReAct exploration | Spider-Agent · Spider 2.0 Snowflake | schema linking F1: 0.35 -> 0.60 | Improved |
| use QAFD-RAG subgraphs with SQL-Agent | Spider-Agent · Spider 2.0 Snowflake | LLM calls: 1482 -> 674 | Improved |
| switch embedding model (small, large, jina-v3, nv-embed-v2, GritLM) (ablation) | QAFD-RAG · UltraDomain Mix | Comprehensiveness: 85.82 to 88.12 | No clear effect |
| vary seed nodes (20-60), mass alpha, and edge weight formula (ablation) | QAFD-RAG · UltraDomain Mix | five GPT-4o dimensions | No clear effect |
| use QAFD-RAG instead of GraphRAG, runtime | QAFD-RAG · UltraDomain Mix | total runtime (s): 50968.59 -> 8948.81 | Improved |
Closest papers by meaning
- Simple is Effective: The Roles of Graphs and Large Language Models in Knowledge-Graph-Based Retrieval-Augmented Generation (ICLR 2025 · similarity 0.91)
- Can Knowledge-Graph-based Retrieval Augmented Generation Really Retrieve What You Need? (NeurIPS 2025 · similarity 0.90)
- RAGraph: A General Retrieval-Augmented Graph Learning Framework (NeurIPS 2024 · similarity 0.89)
- Think-on-Graph 2.0: Deep and Faithful Large Language Model Reasoning with Knowledge-guided Retrieval Augmented Generation (ICLR 2025 · similarity 0.89)
- HyperGraphRAG: Retrieval-Augmented Generation via Hypergraph-Structured Knowledge Representation (NeurIPS 2025 · similarity 0.88)
- GFM-RAG: Graph Foundation Model for Retrieval Augmented Generation (NeurIPS 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
