@inproceedings{wei-etal-2026-cartbench,
title = "{CA}rt{B}ench: A Museum-Grounded Benchmark for Evidence-Faithful {C}hinese Art Understanding",
author = "Wei, Xuefeng and
Zhixuan, Wang and
Zhou, Xuan and
Qu, Zhi and
Li, Hongyao and
Sakai, Yusuke and
Kamigaito, Hidetaka and
Watanabe, Taro",
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.2291/",
pages = "49091--49116",
ISBN = "979-8-89176-462-0",
abstract = "We introduce CARTBENCH, a museum-grounded benchmark for evaluating vision-language models (VLMs) on Chinese artworks beyond short-form recognition and QA. CARTBENCH comprises four subtasks: CURATORQA for evidence-grounded recognition and reasoning, CATALOGCAPTION for structured four-section expert-style appreciation, REINTERPRET for defensible reinterpretation with expert ratings, and CONNOISSEURPAIRS for original-copy discrimination under visually similar confounds. CARTBENCH is built by aligning image-bearing Palace Museum objects from Wikidata with authoritative catalog pages, spanning five art categories across multiple dynasties. Across evaluated VLMs, high overall CURATORQA accuracy can mask sharp drops on evidence-heavy and style-to-period questions; long-form appreciation remains far from expert references; and original-copy discrimination remains near chance. A held-out protocol-replication pilot on 120 artworks from five external institutions reproduces the main task-difficulty pattern, while not preserving the exact overall model ranking."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wei-etal-2026-cartbench">
<titleInfo>
<title>CArtBench: A Museum-Grounded Benchmark for Evidence-Faithful Chinese Art Understanding</title>
</titleInfo>
<name type="personal">
<namePart type="given">Xuefeng</namePart>
<namePart type="family">Wei</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wang</namePart>
<namePart type="family">Zhixuan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xuan</namePart>
<namePart type="family">Zhou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhi</namePart>
<namePart type="family">Qu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hongyao</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yusuke</namePart>
<namePart type="family">Sakai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hidetaka</namePart>
<namePart type="family">Kamigaito</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Taro</namePart>
<namePart type="family">Watanabe</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>We introduce CARTBENCH, a museum-grounded benchmark for evaluating vision-language models (VLMs) on Chinese artworks beyond short-form recognition and QA. CARTBENCH comprises four subtasks: CURATORQA for evidence-grounded recognition and reasoning, CATALOGCAPTION for structured four-section expert-style appreciation, REINTERPRET for defensible reinterpretation with expert ratings, and CONNOISSEURPAIRS for original-copy discrimination under visually similar confounds. CARTBENCH is built by aligning image-bearing Palace Museum objects from Wikidata with authoritative catalog pages, spanning five art categories across multiple dynasties. Across evaluated VLMs, high overall CURATORQA accuracy can mask sharp drops on evidence-heavy and style-to-period questions; long-form appreciation remains far from expert references; and original-copy discrimination remains near chance. A held-out protocol-replication pilot on 120 artworks from five external institutions reproduces the main task-difficulty pattern, while not preserving the exact overall model ranking.</abstract>
<identifier type="citekey">wei-etal-2026-cartbench</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.2291/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>49091</start>
<end>49116</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T CArtBench: A Museum-Grounded Benchmark for Evidence-Faithful Chinese Art Understanding
%A Wei, Xuefeng
%A Zhixuan, Wang
%A Zhou, Xuan
%A Qu, Zhi
%A Li, Hongyao
%A Sakai, Yusuke
%A Kamigaito, Hidetaka
%A Watanabe, Taro
%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 wei-etal-2026-cartbench
%X We introduce CARTBENCH, a museum-grounded benchmark for evaluating vision-language models (VLMs) on Chinese artworks beyond short-form recognition and QA. CARTBENCH comprises four subtasks: CURATORQA for evidence-grounded recognition and reasoning, CATALOGCAPTION for structured four-section expert-style appreciation, REINTERPRET for defensible reinterpretation with expert ratings, and CONNOISSEURPAIRS for original-copy discrimination under visually similar confounds. CARTBENCH is built by aligning image-bearing Palace Museum objects from Wikidata with authoritative catalog pages, spanning five art categories across multiple dynasties. Across evaluated VLMs, high overall CURATORQA accuracy can mask sharp drops on evidence-heavy and style-to-period questions; long-form appreciation remains far from expert references; and original-copy discrimination remains near chance. A held-out protocol-replication pilot on 120 artworks from five external institutions reproduces the main task-difficulty pattern, while not preserving the exact overall model ranking.
%U https://aclanthology.org/2026.findings-emnlp.2291/
%P 49091-49116
Markdown (Informal)
[CArtBench: A Museum-Grounded Benchmark for Evidence-Faithful Chinese Art Understanding](https://aclanthology.org/2026.findings-emnlp.2291/) (Wei et al., Findings 2026)
ACL
- Xuefeng Wei, Wang Zhixuan, Xuan Zhou, Zhi Qu, Hongyao Li, Yusuke Sakai, Hidetaka Kamigaito, and Taro Watanabe. 2026. CArtBench: A Museum-Grounded Benchmark for Evidence-Faithful Chinese Art Understanding. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 49091–49116, Budapest, Hungary. Association for Computational Linguistics.