QUESTER: Query Specification for Generative Retrieval

Arthur Satouf, Yuxuan Zong, Habiboulaye Amadou-Boubacar, Pablo Piantanida, Benjamin Piwowarski
11/7/2025
cs.IRcs.CLcs.LG

No official code found. Best implementation: arthur-75/quester (0 stars, Apache-2.0, last commit 2026-03-01).

Abstract

Generative Retrieval (GR) differs from the traditional index-then-retrieve pipeline by storing relevance in model parameters and directly generating document identifiers. However, GR often struggles to generalize and is costly to scale. We introduce QUESTER (QUEry SpecificaTion gEnerative Retrieval), which reframes GR as query specification generation - in this work, a simple keyword query handled by BM25 - using a (small) LLM. The policy is trained using reinforcement learning techniques (GRPO). Across in- and out-of-domain evaluations, we show that our model is more effective than BM25, and competitive with neural IR models, while maintaining a good efficiency

View on arXivView PDF

Code Implementations(1)

arthur-75/questerCommunity74%
Apache-2.0

QueStER: Query Specification for Generative Keyword-Based Retrieval

00Shell, PythonJan 11, 20267 months agoApache-2.0

Similar Papers

Generative Query Reformulation Using Ensemble Prompting, Document Fusion, and Relevance Feedback

83% match2024

Rethinking On-policy Optimization for Query Augmentation

81% match1 repo32025

An Efficient Rubric-based Generative Verifier for Search-Augmented LLMs

81% match2025

Annotation-Free Reinforcement Learning Query Rewriting via Verifiable Search Reward

81% match2025

Lightweight and Direct Document Relevance Optimization for Generative Information Retrieval

81% match1 repo392025

From Matching to Generation: A Survey on Generative Information Retrieval

80% match1 repo2102024

Cite this paper

@article{satouf2025quester,
  title  = {QUESTER: Query Specification for Generative Retrieval},
  author = {Arthur Satouf and Yuxuan Zong and Habiboulaye Amadou-Boubacar and Pablo Piantanida and Benjamin Piwowarski},
  year   = {2025},
  eprint = {2511.05301},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2511.05301v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2511.05301.svg)](https://codeofpaper.com/paper/2511.05301)

Discussion