@inproceedings{si-etal-2026-abdatabench,
title = "{ABDATABENCH}: Evaluating {AI} Agents for Evidence-Grounded Antibody Record Construction",
author = "Si, Weiye and
Qian, Juncheng and
Litong, Liu and
Jin, Meiyan and
Chen, Haotian and
Lu, Pengrui and
Wu, Yunze and
Yu, Junjie and
Ying, Tianlei and
Liu, Pengfei",
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.1255/",
pages = "28594--28614",
ISBN = "979-8-89176-459-0",
abstract = "AI agents promise to turn scattered scientific evidence into AI-ready datasets, yet existing benchmarks examine isolated NLP skills or paper-level summarization and leave open whether agents can assemble complete, evidence-grounded records across multimodal and heterogeneous scientific sources. To address this gap, we introduce ABDataBench, a general evaluation protocol for antibody functional data construction from scientific literature and patents. ABDataBench contains 44 expert-annotated, OCR-processed documents, including 27 papers and 17 patents, yielding 279 antibody-level records under a unified 22-field schema. We further develop ABCurator, a record-centric multi-agent framework that treats each antibody as the unit for planning, evidence aggregation, enrichment, and validation, coupled with a skill-accumulation workflow that distills expert feedback into reusable agent competencies across evaluation rounds. Across proprietary and open-source models, the strongest configurations achieve scores of 78.3 and 76.0, respectively. Analysis shows that current agents achieve high antibody recall but remain limited by over-generation, sequence attribution errors, and incomplete affinity grounding. These results position antibody record construction as a challenging long-horizon information-extraction task and establish ABDataBench as a benchmark framework for reliable AI-driven scientific data generation. The source code and data for ABDataBench is released at:https://github.com/GAIR-NLP/ABDataBench."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="si-etal-2026-abdatabench">
<titleInfo>
<title>ABDATABENCH: Evaluating AI Agents for Evidence-Grounded Antibody Record Construction</title>
</titleInfo>
<name type="personal">
<namePart type="given">Weiye</namePart>
<namePart type="family">Si</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Juncheng</namePart>
<namePart type="family">Qian</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Liu</namePart>
<namePart type="family">Litong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Meiyan</namePart>
<namePart type="family">Jin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haotian</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pengrui</namePart>
<namePart type="family">Lu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yunze</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Junjie</namePart>
<namePart type="family">Yu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tianlei</namePart>
<namePart type="family">Ying</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pengfei</namePart>
<namePart type="family">Liu</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>AI agents promise to turn scattered scientific evidence into AI-ready datasets, yet existing benchmarks examine isolated NLP skills or paper-level summarization and leave open whether agents can assemble complete, evidence-grounded records across multimodal and heterogeneous scientific sources. To address this gap, we introduce ABDataBench, a general evaluation protocol for antibody functional data construction from scientific literature and patents. ABDataBench contains 44 expert-annotated, OCR-processed documents, including 27 papers and 17 patents, yielding 279 antibody-level records under a unified 22-field schema. We further develop ABCurator, a record-centric multi-agent framework that treats each antibody as the unit for planning, evidence aggregation, enrichment, and validation, coupled with a skill-accumulation workflow that distills expert feedback into reusable agent competencies across evaluation rounds. Across proprietary and open-source models, the strongest configurations achieve scores of 78.3 and 76.0, respectively. Analysis shows that current agents achieve high antibody recall but remain limited by over-generation, sequence attribution errors, and incomplete affinity grounding. These results position antibody record construction as a challenging long-horizon information-extraction task and establish ABDataBench as a benchmark framework for reliable AI-driven scientific data generation. The source code and data for ABDataBench is released at:https://github.com/GAIR-NLP/ABDataBench.</abstract>
<identifier type="citekey">si-etal-2026-abdatabench</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1255/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>28594</start>
<end>28614</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T ABDATABENCH: Evaluating AI Agents for Evidence-Grounded Antibody Record Construction
%A Si, Weiye
%A Qian, Juncheng
%A Litong, Liu
%A Jin, Meiyan
%A Chen, Haotian
%A Lu, Pengrui
%A Wu, Yunze
%A Yu, Junjie
%A Ying, Tianlei
%A Liu, Pengfei
%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 si-etal-2026-abdatabench
%X AI agents promise to turn scattered scientific evidence into AI-ready datasets, yet existing benchmarks examine isolated NLP skills or paper-level summarization and leave open whether agents can assemble complete, evidence-grounded records across multimodal and heterogeneous scientific sources. To address this gap, we introduce ABDataBench, a general evaluation protocol for antibody functional data construction from scientific literature and patents. ABDataBench contains 44 expert-annotated, OCR-processed documents, including 27 papers and 17 patents, yielding 279 antibody-level records under a unified 22-field schema. We further develop ABCurator, a record-centric multi-agent framework that treats each antibody as the unit for planning, evidence aggregation, enrichment, and validation, coupled with a skill-accumulation workflow that distills expert feedback into reusable agent competencies across evaluation rounds. Across proprietary and open-source models, the strongest configurations achieve scores of 78.3 and 76.0, respectively. Analysis shows that current agents achieve high antibody recall but remain limited by over-generation, sequence attribution errors, and incomplete affinity grounding. These results position antibody record construction as a challenging long-horizon information-extraction task and establish ABDataBench as a benchmark framework for reliable AI-driven scientific data generation. The source code and data for ABDataBench is released at:https://github.com/GAIR-NLP/ABDataBench.
%U https://aclanthology.org/2026.emnlp-main.1255/
%P 28594-28614
Markdown (Informal)
[ABDATABENCH: Evaluating AI Agents for Evidence-Grounded Antibody Record Construction](https://aclanthology.org/2026.emnlp-main.1255/) (Si et al., EMNLP 2026)
ACL
- Weiye Si, Juncheng Qian, Liu Litong, Meiyan Jin, Haotian Chen, Pengrui Lu, Yunze Wu, Junjie Yu, Tianlei Ying, and Pengfei Liu. 2026. ABDATABENCH: Evaluating AI Agents for Evidence-Grounded Antibody Record Construction. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 28594–28614, Budapest, Hungary. Association for Computational Linguistics.