@inproceedings{he-etal-2026-sufficiency,
title = "From Sufficiency to Reflection: Reinforcement-Guided Thinking Quality in Retrieval-Augmented Reasoning for {LLM}s",
author = "He, Jie and
Gutierrez Basulto, Victor and
Pan, Jeff Z.",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.213/",
pages = "4883--4908",
ISBN = "979-8-89176-459-0",
abstract = "Reinforcement learning-based retrieval-augmented generation (RAG) methods enhance the reasoning abilities of large language models (LLMs). However, most rely only on final-answer rewards, overlooking intermediate reasoning quality. This paper analyzes existing RAG reasoning models and identifies three main failure patterns: (1) information insufficiency: failure to retrieve adequate support; (2) faulty reasoning: logical or content-level flaws despite sufficient information; (3) answer{--}reasoning inconsistency: a valid reasoning chain with a mismatched final answer. We propose TIRESRAG-R1, a novel framework using a think{--}retrieve{--}reflect process and a multi-dimensional reward system to improve reasoning and stability. TIRESRAG-R1 introduces: (1) a sufficiency reward to encourage thorough retrieval; (2) a reasoning quality reward to assess rationality and accuracy of the reasoning chain; (3) a reflection reward to detect and revise errors. It also employs a difficulty-aware reweighting strategy and training sample filtering to boost performance on complex tasks. Experiments on four multi-hop QA datasets show TIRESRAG-R1 outperforms prior RAG methods and generalizes well to single-hop tasks. Our code is available at https://github.com/probe2/TIRESRAG-R1."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="he-etal-2026-sufficiency">
<titleInfo>
<title>From Sufficiency to Reflection: Reinforcement-Guided Thinking Quality in Retrieval-Augmented Reasoning for LLMs</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jie</namePart>
<namePart type="family">He</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Victor</namePart>
<namePart type="family">Gutierrez Basulto</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jeff</namePart>
<namePart type="given">Z</namePart>
<namePart type="family">Pan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Reinforcement learning-based retrieval-augmented generation (RAG) methods enhance the reasoning abilities of large language models (LLMs). However, most rely only on final-answer rewards, overlooking intermediate reasoning quality. This paper analyzes existing RAG reasoning models and identifies three main failure patterns: (1) information insufficiency: failure to retrieve adequate support; (2) faulty reasoning: logical or content-level flaws despite sufficient information; (3) answer–reasoning inconsistency: a valid reasoning chain with a mismatched final answer. We propose TIRESRAG-R1, a novel framework using a think–retrieve–reflect process and a multi-dimensional reward system to improve reasoning and stability. TIRESRAG-R1 introduces: (1) a sufficiency reward to encourage thorough retrieval; (2) a reasoning quality reward to assess rationality and accuracy of the reasoning chain; (3) a reflection reward to detect and revise errors. It also employs a difficulty-aware reweighting strategy and training sample filtering to boost performance on complex tasks. Experiments on four multi-hop QA datasets show TIRESRAG-R1 outperforms prior RAG methods and generalizes well to single-hop tasks. Our code is available at https://github.com/probe2/TIRESRAG-R1.</abstract>
<identifier type="citekey">he-etal-2026-sufficiency</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.213/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>4883</start>
<end>4908</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T From Sufficiency to Reflection: Reinforcement-Guided Thinking Quality in Retrieval-Augmented Reasoning for LLMs
%A He, Jie
%A Gutierrez Basulto, Victor
%A Pan, Jeff Z.
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F he-etal-2026-sufficiency
%X Reinforcement learning-based retrieval-augmented generation (RAG) methods enhance the reasoning abilities of large language models (LLMs). However, most rely only on final-answer rewards, overlooking intermediate reasoning quality. This paper analyzes existing RAG reasoning models and identifies three main failure patterns: (1) information insufficiency: failure to retrieve adequate support; (2) faulty reasoning: logical or content-level flaws despite sufficient information; (3) answer–reasoning inconsistency: a valid reasoning chain with a mismatched final answer. We propose TIRESRAG-R1, a novel framework using a think–retrieve–reflect process and a multi-dimensional reward system to improve reasoning and stability. TIRESRAG-R1 introduces: (1) a sufficiency reward to encourage thorough retrieval; (2) a reasoning quality reward to assess rationality and accuracy of the reasoning chain; (3) a reflection reward to detect and revise errors. It also employs a difficulty-aware reweighting strategy and training sample filtering to boost performance on complex tasks. Experiments on four multi-hop QA datasets show TIRESRAG-R1 outperforms prior RAG methods and generalizes well to single-hop tasks. Our code is available at https://github.com/probe2/TIRESRAG-R1.
%U https://aclanthology.org/2026.emnlp-main.213/
%P 4883-4908
Markdown (Informal)
[From Sufficiency to Reflection: Reinforcement-Guided Thinking Quality in Retrieval-Augmented Reasoning for LLMs](https://aclanthology.org/2026.emnlp-main.213/) (He et al., EMNLP 2026)
ACL