@inproceedings{guo-etal-2026-anabench,
title = "{A}na{B}ench for Enhancing Scientific Table {\&} Figure Analysis",
author = "Guo, Xuehang and
Lu, Zhiyong and
Hope, Tom and
Wang, Qingyun",
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.1454/",
pages = "33199--33269",
ISBN = "979-8-89176-459-0",
abstract = "In scientific research, analysis requires accurately interpreting complex multimodal knowledge, integrating evidence from different sources, and drawing inferences grounded in domain-specific knowledge. However, current artificial intelligence (AI) systems struggle to consistently demonstrate such capabilities. The complexity and variability of scientific tables and figures, combined with heterogeneous structures and long-context requirements, pose fundamental obstacles to scientific table figure analysis. To quantify these challenges, we introduce AnaBench, a large-scale benchmark featuring 63,178 instances from nine scientific domains, systematically categorized along seven complexity dimensions. To tackle these challenges, we propose Anagent, a multi-agent framework for enhanced scientific table figure analysis through four specialized agents: decomposes tasks into actionable subtasks, retrieves task-specific information through targeted tool execution, synthesizes information to generate coherent analysis, and performs iterative refinement through five-dimensional quality assessment. We further develop modular training strategies that leverage supervised finetuning and specialized reinforcement learning to optimize individual capabilities while maintaining effective collaboration. Comprehensive evaluation across 170 subdomains demonstrates that Anagent achieves substantial improvements, up to $\uparrow13.43$ in training-free settings and $\uparrow42.12$ with finetuning, while revealing that task-oriented reasoning and context-aware problem-solving are essential for high-quality scientific table figure analysis."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="guo-etal-2026-anabench">
<titleInfo>
<title>AnaBench for Enhancing Scientific Table & Figure Analysis</title>
</titleInfo>
<name type="personal">
<namePart type="given">Xuehang</namePart>
<namePart type="family">Guo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhiyong</namePart>
<namePart type="family">Lu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tom</namePart>
<namePart type="family">Hope</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qingyun</namePart>
<namePart type="family">Wang</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>In scientific research, analysis requires accurately interpreting complex multimodal knowledge, integrating evidence from different sources, and drawing inferences grounded in domain-specific knowledge. However, current artificial intelligence (AI) systems struggle to consistently demonstrate such capabilities. The complexity and variability of scientific tables and figures, combined with heterogeneous structures and long-context requirements, pose fundamental obstacles to scientific table figure analysis. To quantify these challenges, we introduce AnaBench, a large-scale benchmark featuring 63,178 instances from nine scientific domains, systematically categorized along seven complexity dimensions. To tackle these challenges, we propose Anagent, a multi-agent framework for enhanced scientific table figure analysis through four specialized agents: decomposes tasks into actionable subtasks, retrieves task-specific information through targeted tool execution, synthesizes information to generate coherent analysis, and performs iterative refinement through five-dimensional quality assessment. We further develop modular training strategies that leverage supervised finetuning and specialized reinforcement learning to optimize individual capabilities while maintaining effective collaboration. Comprehensive evaluation across 170 subdomains demonstrates that Anagent achieves substantial improvements, up to \uparrow13.43 in training-free settings and \uparrow42.12 with finetuning, while revealing that task-oriented reasoning and context-aware problem-solving are essential for high-quality scientific table figure analysis.</abstract>
<identifier type="citekey">guo-etal-2026-anabench</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1454/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>33199</start>
<end>33269</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T AnaBench for Enhancing Scientific Table & Figure Analysis
%A Guo, Xuehang
%A Lu, Zhiyong
%A Hope, Tom
%A Wang, Qingyun
%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 guo-etal-2026-anabench
%X In scientific research, analysis requires accurately interpreting complex multimodal knowledge, integrating evidence from different sources, and drawing inferences grounded in domain-specific knowledge. However, current artificial intelligence (AI) systems struggle to consistently demonstrate such capabilities. The complexity and variability of scientific tables and figures, combined with heterogeneous structures and long-context requirements, pose fundamental obstacles to scientific table figure analysis. To quantify these challenges, we introduce AnaBench, a large-scale benchmark featuring 63,178 instances from nine scientific domains, systematically categorized along seven complexity dimensions. To tackle these challenges, we propose Anagent, a multi-agent framework for enhanced scientific table figure analysis through four specialized agents: decomposes tasks into actionable subtasks, retrieves task-specific information through targeted tool execution, synthesizes information to generate coherent analysis, and performs iterative refinement through five-dimensional quality assessment. We further develop modular training strategies that leverage supervised finetuning and specialized reinforcement learning to optimize individual capabilities while maintaining effective collaboration. Comprehensive evaluation across 170 subdomains demonstrates that Anagent achieves substantial improvements, up to \uparrow13.43 in training-free settings and \uparrow42.12 with finetuning, while revealing that task-oriented reasoning and context-aware problem-solving are essential for high-quality scientific table figure analysis.
%U https://aclanthology.org/2026.emnlp-main.1454/
%P 33199-33269
Markdown (Informal)
[AnaBench for Enhancing Scientific Table & Figure Analysis](https://aclanthology.org/2026.emnlp-main.1454/) (Guo et al., EMNLP 2026)
ACL
- Xuehang Guo, Zhiyong Lu, Tom Hope, and Qingyun Wang. 2026. AnaBench for Enhancing Scientific Table & Figure Analysis. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 33199–33269, Budapest, Hungary. Association for Computational Linguistics.