EUVIMEDEUVIMED entdecken
Crossref

Direct Retrieval-augmented Optimization: Synergizing Knowledge Selection and Language Models

Zhengliang Shi, Lingyong Yan, Weiwei Sun, Yue Feng, Pengjie Ren, Xinyu Ma, Shuaiqiang Wang, Dawei Yin, Maarten de Rijke, Zhaochun Ren

ACM Transactions on Information Systems · 31.05.2026 · 44 (4) · 1-30

DOI: 10.1145/3795527 ↗

Kurz gefasst

Fachartikel in der Zeitschrift ACM Transactions on Information Systems, online erschienen am 11.05.2026. Verfasst von Zhengliang Shi und 9 weiteren Autor:innen. Die Arbeit wurde bisher 4-mal zitiert (Crossref) und nennt 76 Literaturangaben. Gefördert von: National Key Research and Development Program of China, Dutch Research Council, European Union’s Horizon Europe program.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Retrieval-augmented Generation (RAG) integrates Large Language Models (LLMs) with retrievers to access external knowledge, improving the factuality of LLM generation in knowledge-grounded tasks. To optimize the RAG performance, most previous work independently fine-tunes the retriever to adapt to frozen LLMs or trains the LLMs to use documents retrieved by off-the-shelf retrievers, lacking end-to-end training supervision. Recent work addresses this limitation by jointly training these two components but relies on overly simplifying assumptions of document independence, which has been criticized for being far from real-world scenarios. Thus, effectively optimizing the overall RAG performance remains a critical challenge. We propose a Direct Retrieval-augmented Optimization ( DRO ) framework that enables end-to-end training of two key components: (i) a generative knowledge selection model and (ii) an LLM generator. DRO alternates between two phases: (i) document permutation estimation and (ii) re-weighted maximization, progressively improving RAG components through a variational approach. In the estimation step, we treat document permutation as a latent variable and directly estimate its distribution from the selection model by applying an importance sampling strategy. In the maximization step, we calibrate the optimization expectation using importance weights and jointly train the selection model and LLM generator. Our theoretical analysis reveals that DRO is analogous to policy-gradient methods in reinforcement learning. Extensive experiments conducted on five datasets illustrate that DRO outperforms the best baseline with 5–15% improvements in EM and F1. We also qualitatively analyze the stability, convergence, and variance of DRO. (Code is available on DRO GitHub ).

Förderung

  • National Key Research and Development Program of China
  • Dutch Research Council
  • European Union’s Horizon Europe program

Zitieren

APA

Shi, Z., Yan, L., Sun, W., Feng, Y., Ren, P., Ma, X., Wang, S., Yin, D., Rijke, M. D., & Ren, Z. (2026). Direct Retrieval-augmented Optimization: Synergizing Knowledge Selection and Language Models. ACM Transactions on Information Systems, 44(4), 1-30. https://doi.org/10.1145/3795527

Vancouver

Shi Z, Yan L, Sun W, Feng Y, Ren P, Ma X, et al. Direct Retrieval-augmented Optimization: Synergizing Knowledge Selection and Language Models. ACM Transactions on Information Systems. 2026;44(4):1-30. doi:10.1145/3795527

BibTeX
@article{Shi2026,
  author = {Shi, Zhengliang and Yan, Lingyong and Sun, Weiwei and Feng, Yue and Ren, Pengjie and Ma, Xinyu and Wang, Shuaiqiang and Yin, Dawei and Rijke, Maarten de and Ren, Zhaochun},
  title = {Direct Retrieval-augmented Optimization: Synergizing Knowledge Selection and Language Models},
  journal = {ACM Transactions on Information Systems},
  year = {2026},
  volume = {44},
  number = {4},
  pages = {1--30},
  doi = {10.1145/3795527}
}
Kostenlos starten →