@inproceedings{fu-etal-2026-papergraph,
title = "{P}aper{G}raph-Bench: Graph-Guided Benchmark Construction and Multi-Turn Evaluation for Scientific Paper Reading",
author = "Fu, Rao and
Dai, Qing and
Ou-yang, Kaitai and
He, Kang and
Li, Fei and
Ji, Donghong and
Teng, Chong",
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.804/",
doi = "10.18653/2026.findings-emnlp.804",
pages = "17028--17055",
ISBN = "979-8-89176-462-0",
abstract = "Scientific-paper reading requires models to recover a paper{'}s core argument, connect claims to textual and multimodal evidence, and avoid unsupported inferences across dialogue turns. We introduce \textsc{PaperGraph-Bench}, a graph-guided benchmark that turns each paper into a paper-specific evaluation state graph with macro stages, evidence-grounded knowledge components (KCs), verified reasoning edges, multimodal assets, rubrics, and trajectory state. The graph drives macro coverage questions, challenge questions, reasoning threads, and bounded repair turns, while the evaluator tracks KC coverage, challenge failures, hallucination events, and repair outcomes. Experiments on a human-audited prototype with ten models show that coverage, robustness, hallucination type, and repair behavior reveal distinct paper-reading abilities. Our code and current dataset are available at https://github.com/furao0011/PaperGraphBench."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="fu-etal-2026-papergraph">
<titleInfo>
<title>PaperGraph-Bench: Graph-Guided Benchmark Construction and Multi-Turn Evaluation for Scientific Paper Reading</title>
</titleInfo>
<name type="personal">
<namePart type="given">Rao</namePart>
<namePart type="family">Fu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qing</namePart>
<namePart type="family">Dai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kaitai</namePart>
<namePart type="family">Ou-yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kang</namePart>
<namePart type="family">He</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Fei</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Donghong</namePart>
<namePart type="family">Ji</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chong</namePart>
<namePart type="family">Teng</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>Scientific-paper reading requires models to recover a paper’s core argument, connect claims to textual and multimodal evidence, and avoid unsupported inferences across dialogue turns. We introduce PaperGraph-Bench, a graph-guided benchmark that turns each paper into a paper-specific evaluation state graph with macro stages, evidence-grounded knowledge components (KCs), verified reasoning edges, multimodal assets, rubrics, and trajectory state. The graph drives macro coverage questions, challenge questions, reasoning threads, and bounded repair turns, while the evaluator tracks KC coverage, challenge failures, hallucination events, and repair outcomes. Experiments on a human-audited prototype with ten models show that coverage, robustness, hallucination type, and repair behavior reveal distinct paper-reading abilities. Our code and current dataset are available at https://github.com/furao0011/PaperGraphBench.</abstract>
<identifier type="citekey">fu-etal-2026-papergraph</identifier>
<identifier type="doi">10.18653/2026.findings-emnlp.804</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.804/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>17028</start>
<end>17055</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T PaperGraph-Bench: Graph-Guided Benchmark Construction and Multi-Turn Evaluation for Scientific Paper Reading
%A Fu, Rao
%A Dai, Qing
%A Ou-yang, Kaitai
%A He, Kang
%A Li, Fei
%A Ji, Donghong
%A Teng, Chong
%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 fu-etal-2026-papergraph
%X Scientific-paper reading requires models to recover a paper’s core argument, connect claims to textual and multimodal evidence, and avoid unsupported inferences across dialogue turns. We introduce PaperGraph-Bench, a graph-guided benchmark that turns each paper into a paper-specific evaluation state graph with macro stages, evidence-grounded knowledge components (KCs), verified reasoning edges, multimodal assets, rubrics, and trajectory state. The graph drives macro coverage questions, challenge questions, reasoning threads, and bounded repair turns, while the evaluator tracks KC coverage, challenge failures, hallucination events, and repair outcomes. Experiments on a human-audited prototype with ten models show that coverage, robustness, hallucination type, and repair behavior reveal distinct paper-reading abilities. Our code and current dataset are available at https://github.com/furao0011/PaperGraphBench.
%R 10.18653/2026.findings-emnlp.804
%U https://aclanthology.org/2026.findings-emnlp.804/
%U https://doi.org/10.18653/2026.findings-emnlp.804
%P 17028-17055
Markdown (Informal)
[PaperGraph-Bench: Graph-Guided Benchmark Construction and Multi-Turn Evaluation for Scientific Paper Reading](https://aclanthology.org/2026.findings-emnlp.804/) (Fu et al., Findings 2026)
ACL