@inproceedings{gupta-etal-2025-advsumm,
title = "{A}dv{S}umm: Adversarial Training for Bias Mitigation in Text Summarization",
author = "Gupta, Mukur and
Varimalla, Nikhil Reddy and
Deas, Nicholas and
Subbiah, Melanie and
McKeown, Kathleen",
editor = "Dong, Yue and
Xiao, Wen and
Zhang, Haopeng and
Zhang, Rui and
Ernst, Ori and
Wang, Lu and
Liu, Fei",
booktitle = "Proceedings of The 5th New Frontiers in Summarization Workshop",
month = nov,
year = "2025",
address = "Hybrid",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.newsum-main.12/",
pages = "172--182",
ISBN = "979-8-89176-337-1",
abstract = "Large Language Models (LLMs) have achieved impressive performance in text summarization and are increasingly deployed in real-world applications. However, these systems often inherit associative and framing biases from pre-training data, leading to inappropriate or unfair outputs in downstream tasks. In this work, we present AdvSumm (Adversarial Summarization), a domain-agnostic training framework designed to mitigate bias in text summarization through improved generalization. Inspired by adversarial robustness, AdvSumm introduces a novel Perturber component that applies gradient-guided perturbations at the embedding level of Sequence-to-Sequence models, enhancing the model{'}s robustness to input variations. We empirically demonstrate that AdvSumm effectively reduces different types of bias in summarization{---}specifically, name-nationality bias and political framing bias{---}without compromising summarization quality. Compared to standard transformers and data augmentation techniques like back-translation, AdvSumm achieves stronger bias mitigation performance across benchmark datasets."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="gupta-etal-2025-advsumm">
<titleInfo>
<title>AdvSumm: Adversarial Training for Bias Mitigation in Text Summarization</title>
</titleInfo>
<name type="personal">
<namePart type="given">Mukur</namePart>
<namePart type="family">Gupta</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nikhil</namePart>
<namePart type="given">Reddy</namePart>
<namePart type="family">Varimalla</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nicholas</namePart>
<namePart type="family">Deas</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Melanie</namePart>
<namePart type="family">Subbiah</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kathleen</namePart>
<namePart type="family">McKeown</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of The 5th New Frontiers in Summarization Workshop</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yue</namePart>
<namePart type="family">Dong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wen</namePart>
<namePart type="family">Xiao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haopeng</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rui</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ori</namePart>
<namePart type="family">Ernst</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lu</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Fei</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Hybrid</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-337-1</identifier>
</relatedItem>
<abstract>Large Language Models (LLMs) have achieved impressive performance in text summarization and are increasingly deployed in real-world applications. However, these systems often inherit associative and framing biases from pre-training data, leading to inappropriate or unfair outputs in downstream tasks. In this work, we present AdvSumm (Adversarial Summarization), a domain-agnostic training framework designed to mitigate bias in text summarization through improved generalization. Inspired by adversarial robustness, AdvSumm introduces a novel Perturber component that applies gradient-guided perturbations at the embedding level of Sequence-to-Sequence models, enhancing the model’s robustness to input variations. We empirically demonstrate that AdvSumm effectively reduces different types of bias in summarization—specifically, name-nationality bias and political framing bias—without compromising summarization quality. Compared to standard transformers and data augmentation techniques like back-translation, AdvSumm achieves stronger bias mitigation performance across benchmark datasets.</abstract>
<identifier type="citekey">gupta-etal-2025-advsumm</identifier>
<location>
<url>https://aclanthology.org/2025.newsum-main.12/</url>
</location>
<part>
<date>2025-11</date>
<extent unit="page">
<start>172</start>
<end>182</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T AdvSumm: Adversarial Training for Bias Mitigation in Text Summarization
%A Gupta, Mukur
%A Varimalla, Nikhil Reddy
%A Deas, Nicholas
%A Subbiah, Melanie
%A McKeown, Kathleen
%Y Dong, Yue
%Y Xiao, Wen
%Y Zhang, Haopeng
%Y Zhang, Rui
%Y Ernst, Ori
%Y Wang, Lu
%Y Liu, Fei
%S Proceedings of The 5th New Frontiers in Summarization Workshop
%D 2025
%8 November
%I Association for Computational Linguistics
%C Hybrid
%@ 979-8-89176-337-1
%F gupta-etal-2025-advsumm
%X Large Language Models (LLMs) have achieved impressive performance in text summarization and are increasingly deployed in real-world applications. However, these systems often inherit associative and framing biases from pre-training data, leading to inappropriate or unfair outputs in downstream tasks. In this work, we present AdvSumm (Adversarial Summarization), a domain-agnostic training framework designed to mitigate bias in text summarization through improved generalization. Inspired by adversarial robustness, AdvSumm introduces a novel Perturber component that applies gradient-guided perturbations at the embedding level of Sequence-to-Sequence models, enhancing the model’s robustness to input variations. We empirically demonstrate that AdvSumm effectively reduces different types of bias in summarization—specifically, name-nationality bias and political framing bias—without compromising summarization quality. Compared to standard transformers and data augmentation techniques like back-translation, AdvSumm achieves stronger bias mitigation performance across benchmark datasets.
%U https://aclanthology.org/2025.newsum-main.12/
%P 172-182
Markdown (Informal)
[AdvSumm: Adversarial Training for Bias Mitigation in Text Summarization](https://aclanthology.org/2025.newsum-main.12/) (Gupta et al., NewSum 2025)
ACL