@inproceedings{kirsch-etal-2025-pm3,
title = "{PM}3-{KIE}: A Probabilistic Multi-Task Meta-Model for Document Key Information Extraction",
author = "Kirsch, Birgit and
Allende-Cid, H{\'e}ctor and
Rueping, Stefan",
editor = "Che, Wanxiang and
Nabende, Joyce and
Shutova, Ekaterina and
Pilehvar, Mohammad Taher",
booktitle = "Findings of the Association for Computational Linguistics: ACL 2025",
month = jul,
year = "2025",
address = "Vienna, Austria",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.findings-acl.1075/",
doi = "10.18653/v1/2025.findings-acl.1075",
pages = "20890--20912",
ISBN = "979-8-89176-256-5",
abstract = "Key Information Extraction (KIE) from visually rich documents is commonly approached as either fine-grained token classification or coarse-grained entity extraction. While token-level models capture spatial and visual cues, entity-level models better represent logical dependencies and align with real-world use cases.We introduce PM3-KIE, a probabilistic multi-task meta-model that incorporates both fine-grained and coarse-grained models. It serves as a lightweight reasoning layer that jointly predicts entities and all appearances in a document. PM3-KIE incorporates domain-specific schema constraints to enforce logical consistency and integrates large language models for semantic validation, thereby reducing extraction errors.Experiments on two public datasets, DeepForm and FARA, show that PM3-KIE outperforms three state-of-the-art models and a stacked ensemble, achieving a statistically significant 2{\%} improvement in F1 score."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="kirsch-etal-2025-pm3">
<titleInfo>
<title>PM3-KIE: A Probabilistic Multi-Task Meta-Model for Document Key Information Extraction</title>
</titleInfo>
<name type="personal">
<namePart type="given">Birgit</namePart>
<namePart type="family">Kirsch</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Héctor</namePart>
<namePart type="family">Allende-Cid</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Stefan</namePart>
<namePart type="family">Rueping</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: ACL 2025</title>
</titleInfo>
<name type="personal">
<namePart type="given">Wanxiang</namePart>
<namePart type="family">Che</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Joyce</namePart>
<namePart type="family">Nabende</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ekaterina</namePart>
<namePart type="family">Shutova</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mohammad</namePart>
<namePart type="given">Taher</namePart>
<namePart type="family">Pilehvar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Vienna, Austria</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-256-5</identifier>
</relatedItem>
<abstract>Key Information Extraction (KIE) from visually rich documents is commonly approached as either fine-grained token classification or coarse-grained entity extraction. While token-level models capture spatial and visual cues, entity-level models better represent logical dependencies and align with real-world use cases.We introduce PM3-KIE, a probabilistic multi-task meta-model that incorporates both fine-grained and coarse-grained models. It serves as a lightweight reasoning layer that jointly predicts entities and all appearances in a document. PM3-KIE incorporates domain-specific schema constraints to enforce logical consistency and integrates large language models for semantic validation, thereby reducing extraction errors.Experiments on two public datasets, DeepForm and FARA, show that PM3-KIE outperforms three state-of-the-art models and a stacked ensemble, achieving a statistically significant 2% improvement in F1 score.</abstract>
<identifier type="citekey">kirsch-etal-2025-pm3</identifier>
<identifier type="doi">10.18653/v1/2025.findings-acl.1075</identifier>
<location>
<url>https://aclanthology.org/2025.findings-acl.1075/</url>
</location>
<part>
<date>2025-07</date>
<extent unit="page">
<start>20890</start>
<end>20912</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T PM3-KIE: A Probabilistic Multi-Task Meta-Model for Document Key Information Extraction
%A Kirsch, Birgit
%A Allende-Cid, Héctor
%A Rueping, Stefan
%Y Che, Wanxiang
%Y Nabende, Joyce
%Y Shutova, Ekaterina
%Y Pilehvar, Mohammad Taher
%S Findings of the Association for Computational Linguistics: ACL 2025
%D 2025
%8 July
%I Association for Computational Linguistics
%C Vienna, Austria
%@ 979-8-89176-256-5
%F kirsch-etal-2025-pm3
%X Key Information Extraction (KIE) from visually rich documents is commonly approached as either fine-grained token classification or coarse-grained entity extraction. While token-level models capture spatial and visual cues, entity-level models better represent logical dependencies and align with real-world use cases.We introduce PM3-KIE, a probabilistic multi-task meta-model that incorporates both fine-grained and coarse-grained models. It serves as a lightweight reasoning layer that jointly predicts entities and all appearances in a document. PM3-KIE incorporates domain-specific schema constraints to enforce logical consistency and integrates large language models for semantic validation, thereby reducing extraction errors.Experiments on two public datasets, DeepForm and FARA, show that PM3-KIE outperforms three state-of-the-art models and a stacked ensemble, achieving a statistically significant 2% improvement in F1 score.
%R 10.18653/v1/2025.findings-acl.1075
%U https://aclanthology.org/2025.findings-acl.1075/
%U https://doi.org/10.18653/v1/2025.findings-acl.1075
%P 20890-20912
Markdown (Informal)
[PM3-KIE: A Probabilistic Multi-Task Meta-Model for Document Key Information Extraction](https://aclanthology.org/2025.findings-acl.1075/) (Kirsch et al., Findings 2025)
ACL