Sage Journals: Discover world-class research

Abstract

Retrieval-Augmented Generation (RAG) enhances language model responses by incorporating external knowledge. However, its effectiveness heavily depends on the quality of the retrieved documents. Using a fixed number of retrieved documents $K$ often fails to adapt to varying query complexity, leading either to irrelevant retrievals or to missing crucial evidence. To address this issue, we propose DyRAG, a hybrid retrieval framework that dynamically adjusts $K$ based on query characteristics while maintaining computational efficiency. Our method improves retrieval performance by maximizing relevant information and minimizing noise. We evaluate DyRAG across recommendation, question answering, and fact-checking tasks, where it consistently outperforms fixed- $K$ and hybrid baselines. Compared to traditional approaches, DyRAG demonstrates greater robustness and adaptability across diverse domains.

Keywords

retrieval-augmented generation (RAG)dynamic retrieval (dyRAG)large language model query complexity relevant information

Get full access to this article

View all access options for this article.

References

Alshuaibi

Almaayah

Ali

(2025). Machine learning for cybersecurity issues: A systematic review. Journal of Cyber Security and Risk Auditing, 2025(1), 36–46.

Bai

Wang

Zhang

Shang

Wang

Liu

(2020). Sparterm: Learning term-based sparse representation for fast text retrieval. CoRR abs/2010.00768. https://arxiv.org/abs/2010.00768

Brown

T. B.

Mann

Ryder

Subbiah

Kaplan

Dhariwal

Neelakantan

Shyam

Sastry

Askell

(2020). Language models are few-shot learners. CoRR abs/2005.14165. https://arxiv.org/abs/2005.14165

Chen

Tworek

Jun

Yuan

de Oliveira Pinto

H. P.

Kaplan

Edwards

Burda

Joseph

Brockman

Ray

(2021). Evaluating large language models trained on code. CoRR abs/2107.03374. https://arxiv.org/abs/2107.03374

Chuyuan

Shengda

Hongchun

Shuqiang

Jie

(2025). Enhanced recommendation systems with retrieval-augmented large language model. JAIR, 82, 1147–1173. https://doi.org/10.1613/jair.1.17809

Culpepper

J. S.

Clarke

C. L. A.

Lin

(2016). Dynamic cutoff prediction in multi-stage retrieval systems. In Proceedings of the 21st Australasian document computing symposium, Melbourne, Australia. https://culpepper.io/publications/ccl16-adcs.pdf

Devlin

Chang

M. W.

Lee

Toutanova

(2019). BERT: Pre-training of deep bidirectional transformers for language understanding. In J. Burstein, C. Doran & T. Solorio (Eds.), Proceedings of the 2019 conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Minneapolis, Minnesota. https://doi.org/10.18653/v1/N19-1423

Ding

Zhai

(2024). Efficient retrieval with learned similarities. CoRR abs/2407.15462. https://doi.org/10.48550/ARXIV.2407.15462

Farea

Al-Sarem

Saeed

Al-Hadhrami

Alshammari

(2022). Question answering systems: A review on present developments, challenges and future trends. IEEE Access, 10, 91171–91191. https://doi.org/10.1109/ACCESS.2022.3201371

10.

Gao

Xiong

Gao

Jia

Pan

Dai

Sun

Guo

Wang

(2023). Retrieval-augmented generation for large language models: A survey. CoRR abs/2312.10997. https://doi.org/10.48550/arXiv.2312.10997

11.

Wang

Cho

V. O.

(2018). Search engine guided neural machine translation. In Proceedings of the AAAI conference on artificial intelligence (Vol. 32, pp. 5133–5140). https://doi.org/10.1609/AAAI.V32I1.12013

12.

Gupta

Ranjan

Singh

S. N.

(2024). A comprehensive survey of retrieval-augmented generation (RAG): Evolution, current landscape and future directions. CoRR abs/2410.12837. https://doi.org/10.48550/ARXIV.2410.12837

13.

Harper

F. M.

Konstan

J. A.

(2015). The movielens datasets: History and context. ACM Transactions on Interactive Intelligent Systems, 5(4), 1–19. https://doi.org/10.1145/2827872

14.

Hei

Liu

Qiao

Jiao

Song

Tian

Lin

(2024). DR-RAG: Applying dynamic document relevance to retrieval-augmented generation for question-answering. CoRR abs/2406.07348. https://doi.org/10.48550/ARXIV.2406.07348

15.

Howard

Ruder

(2018). Universal language model fine-tuning for text classification. In Proceedings of the 56th annual meeting of the Association for Computational Linguistics, ACL , Melbourne, Australia (pp. 328–339). https://doi.org/10.18653/V1/P18-1031

16.

Huang

(2024). A survey on retrieval-augmented text generation for large language models. CoRR abs/2404.10981. https://doi.org/10.48550/ARXIV.2404.10981

17.

Hulett

Hall

(2012). Dynamic selection of k nearest neighbors in instance-based learning. In 2012 IEEE 13th international conference on information reuse & integration (IRI), Las Vegas, NV, USA. https://doi.org/10.1109/IRI.2012.6302995

18.

Jadon

Patil

(2023). A comprehensive survey of evaluation techniques for recommendation systems. CoRR abs/2312.16015. https://doi.org/10.48550/ARXIV.2312.16015

19.

Kang

McAuley

J. J.

(2018). Self-attentive sequential recommendation. CoRR abs/1808.09781. http://arxiv.org/abs/1808.09781

20.

Karpukhin

Oguz

Min

Edunov

Chen

Yih

(2020). Dense passage retrieval for open-domain question answering. CoRR abs/2004.04906. https://arxiv.org/abs/2004.04906

21.

Kumar

Pawar

P. P.

Ananthan

Indhumathi

Murugan

M. S.

(2024). CHOS_LSTM: Chebyshev osprey optimization-based model for detecting attacks. In 2024 3rd International conference on artificial intelligence for internet of things (AIIoT) (pp. 1–6). IEEE. https://doi.org/10.1109/AIIoT61547.2024.10569818

22.

Lafferty

Zhai

(2017). Document language models, query models, and risk minimization for information retrieval. SIGIR Forum, 51(2), 251–259. https://doi.org/10.1145/3130348.3130375

23.

Lewis

Liu

Goyal

Ghazvininejad

Mohamed

Levy

Stoyanov

Zettlemoyer

(2020). BART: Denoising sequence-to-sequence pre-training for natural language generation, translation, and comprehension (pp. 7871–7880). https://doi.org/10.18653/V1/2020.ACL-MAIN.703

24.

Lewis

Perez

Piktus

Petroni

Karpukhin

Goyal

Küttler

Lewis

Yih

W. T.

Rocktäschel

Riedel

(2020). Retrieval-augmented generation for knowledge-intensive NLP tasks. CoRR abs/2005.11401. https://arxiv.org/abs/2005.11401

25.

Liang

Wang

Song

Yang

Niu

Liu

Yao

Xiong

(2024). Controllable text generation for large language models: A survey. CoRR abs/2408.12599. https://doi.org/10.48550/ARXIV.2408.12599

26.

Matoušová

Trojovský

Dehghani

Trojovská

Kostra

(2023). Mother optimization algorithm: A new human-based metaheuristic approach for solving engineering optimization. Scientific Reports, 13, 11616. https://doi.org/10.1038/s41598-023-38571-3

27.

Mousavi

S. M.

Alghisi

Riccardi

(2025). Llms as repositories of factual knowledge: Limitations and solutions. CoRR abs/2501.12774. https://doi.org/10.48550/ARXIV.2501.12774

28.

Nadimi-Shahraki

M. I. H.

Mirjalili

Gandomi

A. H.

(2023). Kookaburra optimization algorithm: A new bio-inspired metaheuristic algorithm for solving optimization problems. Biomimetics 8(6), 470. https://doi.org/10.3390/biomimetics8060470

29.

Naveed

Khan

A. U.

Qiu

Saqib

Anwar

Usman

Barnes

Mian

(2023). A comprehensive overview of large language models. CoRR abs/2307.06435. https://doi.org/10.48550/arXiv.2307.06435

30.

Radford

Kim

J. W.

Hallacy

Ramesh

Goh

Agarwal

Sastry

Askell

Mishkin

Clark

Krueger

Sutskever

(2021). Learning transferable visual models from natural language supervision. CoRR abs/2103.00020. https://arxiv.org/abs/2103.00020

31.

Raffel

Shazeer

Roberts

Lee

Narang

Matena

Zhou

Liu

P. J.

(2020). Exploring the limits of transfer learning with a unified text-to-text transformer. Journal of Machine Learning Research 21, 140:1–140:67. https://arxiv.org/abs/1910.10683

32.

Robertson

S. E.

Walker

(1997). On relevance weights with little relevance information. In N. J. Belkin, A. D. Narasimhalu, P. Willett, W. R. Hersh, F. Can & E. M. Voorhees (Eds.), SIGIR ’97: Proceedings of the 20th annual international ACM SIGIR conference on research and development in information retrieval, Philadelphia, PA, USA (pp. 16–24). https://doi.org/10.1145/258525.258529

33.

Sawarkar

Mangal

Solanki

S. R.

(2024). Blended RAG: Improving RAG (retriever-augmented generation) accuracy with semantic search and hybrid query-based retrievers. CoRR abs/2404.07220. https://doi.org/10.48550/ARXIV.2404.07220

34.

Thakur

Reimers

Rücklé

Srivastava

Gurevych

(2021). Beir: A heterogeneous benchmark for zero-shot evaluation of information retrieval models. In Proceedings of the thirty-fifth conference on neural information processing systems (NeurIPS 2021) datasets and benchmarks track (Round 2).

35.

Vaswani

Shazeer

Parmar

Uszkoreit

Jones

Gomez

A. N.

Kaiser

Polosukhin

(2017). Attention is all you need. In Advances in neural information processing systems (Vol. 30). https://arxiv.org/abs/1706.03762

36.

Yuan

Liu

Yuan

Sun

Zhang

(2024). A hybrid RAG system with comprehensive enhancement on complex reasoning. CoRR abs/2408.05141. https://doi.org/10.48550/ARXIV.2408.05141

37.

Zaragoza

Robertson

S. E.

(2009). The probabilistic relevance framework: BM25 and beyond. Foundations and Trends® in Information Retrieval, 3(4), 333–389. https://doi.org/10.1561/1500000019

38.

Zhai

Gong

Wang

Sun

Yan

Liu

(2023). Revisiting neural retrieval on accelerators. CoRR abs/2306.04039. https://doi.org/10.48550/ARXIV.2306.04039

39.

Zhang

Chen

Cao

Chen

Cohn

Fang

(2023). A survey for efficient open domain question answering. In A. Rogers, J. B. Graber & N. Okazaki (Eds.), Proceedings of the 61st annual meeting of the association for computational linguistics (Volume 1: Long Papers), Toronto, Canada (pp. 14447–14465). https://doi.org/10.18653/v1/2023.acl-long.808

40.

Zhao

Zhang

Wang

Geng

Yang

Zhang

Cui

(2024). Retrieval-augmented generation for ai-generated content: A survey. CoRR abs/2402.19473. https://doi.org/10.48550/ARXIV.2402.19473

41.

Zhou

Agichtein

(2020). Rlirank: Learning to rank with reinforcement learning for dynamic search. In Y. Huang, I. King, T. Liu & M. van Steen (eds.) WWW ’20: The web conference 2020, Taipei, Taiwan, April 20-24 (pp. 2842–2848). https://doi.org/10.1145/3366423.3380047

42.

Zhu

Shimada

Taniguchi

Ohkuma

(2024). STAYKATE: Hybrid in-context example selection combining representativeness sampling and retrieval-based approach - A case study on science domains. CoRR abs/2412.20043. https://doi.org/10.48550/ARXIV.2412.20043

43.

Zhu

Zhang

Gai

(2018). Learning tree-based deep model for recommender systems. CoRR abs/1801.02294. http://arxiv.org/abs/1801.02294

44.

Zhuang

Duan

Zhu

Xiong

(2021). A comprehensive survey on transfer learning. Proceedings of the IEEE, 109(1), 43–76.

Boosting RAG Efficiency With dyRAG: Dynamic Candidate Selection for Optimal Retrieval

Abstract

Keywords

Get full access to this article

References