@inproceedings{wang-etal-2026-sdrag,
title = "{SDRAG}: Sub-question Dependency-Aware Retrieval-Augmented Generation for Multi-hop Question Answering",
author = "Wang, Mengze and
Liu, Huan and
Dong, Yixiang and
Zhang, Lingling and
Hu, Bo and
Li, Zhuohang",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.1468/",
pages = "31248--31264",
ISBN = "979-8-89176-462-0",
abstract = "Multi-hop question answering (QA) requires integrating evidence across multiple documents and reasoning over interdependent intermediate steps, posing significant challenges for retrieval-augmented generation (RAG) systems. While recent RAG-based approaches interleave retrieval and reasoning to improve factual grounding, they often struggle with complex multi-hop questions due to retrieval{--}reason mismatch, ignored dependencies among sub-questions, and error propagation during decomposition. In this work, we propose SDRAG, a sub-question dependency-aware RAG framework that explicitly models the intrinsic logic of multi-hop questions as a dependency graph. SDRAG first decomposes the question into fine-grained sub-questions and explicitly models their dependency relations, then refines the graph by pruning unreliable or redundant nodes, and finally traverses the pruned graph to coordinate retrieval and reasoning. To improve graph construction quality, we further introduce a supervised fine-tuning strategy that independently trains task-specific LoRA adapters for decomposition, dependency prediction, and pruning. Experiments on three common multi-hop QA benchmarks show that SDRAG consistently outperforms strong RAG-based baselines, while additional analyses demonstrate the effectiveness of dependency modeling, graph pruning, and fine-tuning."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wang-etal-2026-sdrag">
<titleInfo>
<title>SDRAG: Sub-question Dependency-Aware Retrieval-Augmented Generation for Multi-hop Question Answering</title>
</titleInfo>
<name type="personal">
<namePart type="given">Mengze</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Huan</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yixiang</namePart>
<namePart type="family">Dong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lingling</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bo</namePart>
<namePart type="family">Hu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhuohang</namePart>
<namePart type="family">Li</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>Findings of the Association for Computational Linguistics: EMNLP 2026</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">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Multi-hop question answering (QA) requires integrating evidence across multiple documents and reasoning over interdependent intermediate steps, posing significant challenges for retrieval-augmented generation (RAG) systems. While recent RAG-based approaches interleave retrieval and reasoning to improve factual grounding, they often struggle with complex multi-hop questions due to retrieval–reason mismatch, ignored dependencies among sub-questions, and error propagation during decomposition. In this work, we propose SDRAG, a sub-question dependency-aware RAG framework that explicitly models the intrinsic logic of multi-hop questions as a dependency graph. SDRAG first decomposes the question into fine-grained sub-questions and explicitly models their dependency relations, then refines the graph by pruning unreliable or redundant nodes, and finally traverses the pruned graph to coordinate retrieval and reasoning. To improve graph construction quality, we further introduce a supervised fine-tuning strategy that independently trains task-specific LoRA adapters for decomposition, dependency prediction, and pruning. Experiments on three common multi-hop QA benchmarks show that SDRAG consistently outperforms strong RAG-based baselines, while additional analyses demonstrate the effectiveness of dependency modeling, graph pruning, and fine-tuning.</abstract>
<identifier type="citekey">wang-etal-2026-sdrag</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1468/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>31248</start>
<end>31264</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T SDRAG: Sub-question Dependency-Aware Retrieval-Augmented Generation for Multi-hop Question Answering
%A Wang, Mengze
%A Liu, Huan
%A Dong, Yixiang
%A Zhang, Lingling
%A Hu, Bo
%A Li, Zhuohang
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F wang-etal-2026-sdrag
%X Multi-hop question answering (QA) requires integrating evidence across multiple documents and reasoning over interdependent intermediate steps, posing significant challenges for retrieval-augmented generation (RAG) systems. While recent RAG-based approaches interleave retrieval and reasoning to improve factual grounding, they often struggle with complex multi-hop questions due to retrieval–reason mismatch, ignored dependencies among sub-questions, and error propagation during decomposition. In this work, we propose SDRAG, a sub-question dependency-aware RAG framework that explicitly models the intrinsic logic of multi-hop questions as a dependency graph. SDRAG first decomposes the question into fine-grained sub-questions and explicitly models their dependency relations, then refines the graph by pruning unreliable or redundant nodes, and finally traverses the pruned graph to coordinate retrieval and reasoning. To improve graph construction quality, we further introduce a supervised fine-tuning strategy that independently trains task-specific LoRA adapters for decomposition, dependency prediction, and pruning. Experiments on three common multi-hop QA benchmarks show that SDRAG consistently outperforms strong RAG-based baselines, while additional analyses demonstrate the effectiveness of dependency modeling, graph pruning, and fine-tuning.
%U https://aclanthology.org/2026.findings-emnlp.1468/
%P 31248-31264
Markdown (Informal)
[SDRAG: Sub-question Dependency-Aware Retrieval-Augmented Generation for Multi-hop Question Answering](https://aclanthology.org/2026.findings-emnlp.1468/) (Wang et al., Findings 2026)
ACL