Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Hypothesis-Conditioned Query Rewriting for Decision-Useful Retrieval

Hangeol Chang, Changsun Lee, Seungjoon Rho, Junho Yeo, Jong Chul Ye

Abstract

Retrieval-Augmented Generation (RAG) improves Large Language Models (LLMs) by grounding generation in external, non-parametric knowledge. However, when a task requires choosing among competing options, simply grounding generation in broadly relevant context is often insufficient to drive the final decision. Existing RAG methods typically rely on a single initial query, which often favors topical relevance over decision-relevant evidence, and therefore retrieves background information that can fail to discriminate among answer options. To address this issue, here we propose Hypothesis-Conditioned Query Rewriting (HCQR), a training-free pre-retrieval framework that reorients RAG from topic-oriented retrieval to evidence-oriented retrieval. HCQR first derives a lightweight working hypothesis from the input question and candidate options, and then rewrites retrieval into three targeted queries that seek evidence to: (1) support the hypothesis, (2) distinguish it from competing alternatives, and (3) verify salient clues in the question. This approach enables context retrieval that is more directly aligned with answer selection, allowing the generator to confirm or overturn the initial hypothesis based on the retrieved evidence. Experiments on MedQA and MMLU-Med show that HCQR consistently outperforms single-query RAG and re-rank/filter baselines, improving average accuracy over Simple RAG by 5.9 and 3.6 points, respectively. Code is available at https://anonymous.4open.science/r/HCQR-1C2E.

Hypothesis-Conditioned Query Rewriting for Decision-Useful Retrieval

Abstract

Retrieval-Augmented Generation (RAG) improves Large Language Models (LLMs) by grounding generation in external, non-parametric knowledge. However, when a task requires choosing among competing options, simply grounding generation in broadly relevant context is often insufficient to drive the final decision. Existing RAG methods typically rely on a single initial query, which often favors topical relevance over decision-relevant evidence, and therefore retrieves background information that can fail to discriminate among answer options. To address this issue, here we propose Hypothesis-Conditioned Query Rewriting (HCQR), a training-free pre-retrieval framework that reorients RAG from topic-oriented retrieval to evidence-oriented retrieval. HCQR first derives a lightweight working hypothesis from the input question and candidate options, and then rewrites retrieval into three targeted queries that seek evidence to: (1) support the hypothesis, (2) distinguish it from competing alternatives, and (3) verify salient clues in the question. This approach enables context retrieval that is more directly aligned with answer selection, allowing the generator to confirm or overturn the initial hypothesis based on the retrieved evidence. Experiments on MedQA and MMLU-Med show that HCQR consistently outperforms single-query RAG and re-rank/filter baselines, improving average accuracy over Simple RAG by 5.9 and 3.6 points, respectively. Code is available at https://anonymous.4open.science/r/HCQR-1C2E.
Paper Structure (43 sections, 7 equations, 11 figures, 12 tables)

This paper contains 43 sections, 7 equations, 11 figures, 12 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. RAG design and harmful context.
  4. Query rewriting and query transformation.
  5. Reasoning-guided and agentic retrieval.
  6. Post-retrieval selection and retrieval utility.
  7. Medical RAG as a representative testbed.
  8. Method
  9. Problem Setting
  10. HCQR: Overview
  11. Stage 1: Hypothesis Formulator
  12. Stage 2: Query Rewriter
  13. Retriever, Fusion, and Generation
  14. Experimental Setup
  15. Benchmarks
  16. ...and 28 more sections

Figures (11)

  • Figure 1: HCQR pipeline. A hypothesis formulator proposes a lightweight working hypothesis together with the evidence expected under that hypothesis. The query rewriter turns this state into three verification-oriented queries. The working hypothesis steers retrieval but is not shown to the generator by default.
  • Figure 2: Prompt used for the context-utility judge.
  • Figure 3: Prompt used for the no-retrieval CoT baseline.
  • Figure 6: Prompt used for generating hypothetical documents in the HyDE baseline.
  • Figure 9: Prompt used for the HCQR hypothesis-formulation stage.
  • ...and 6 more figures