<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="haque-etal-2026-fair">
    <titleInfo>
        <title>Fair RAG: End-to-End Fairness Across Retrieval and Generation</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Farsheed</namePart>
        <namePart type="family">Haque</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Zhe</namePart>
        <namePart type="family">Fu</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Ramit</namePart>
        <namePart type="family">Aditya</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Depeng</namePart>
        <namePart type="family">Xu</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Xi</namePart>
        <namePart type="family">Niu</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2026-07</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Findings of the Association for Computational Linguistics: ACL 2026</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Maria</namePart>
            <namePart type="family">Liakata</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Viviane</namePart>
            <namePart type="given">P</namePart>
            <namePart type="family">Moreira</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Jiajun</namePart>
            <namePart type="family">Zhang</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">David</namePart>
            <namePart type="family">Jurgens</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">San Diego, California, United States</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-395-1</identifier>
    </relatedItem>
    <abstract>Large Language Models (LLMs) used in Retrieval-Augmented Generation (RAG) can amplify demographic bias: retrievers may surface skewed context and generators can propagate that skew into decisions. Prior work typically treats fairness in retrieval or generation in isolation, leaving end-to-end fairness in RAG underexplored. We propose a post-hoc pipeline that jointly controls both stages: (i) a Fair Greedy Reranker (FGR) that builds prefix-balanced slates toward a target group mix; (ii) a Residual Slate Bias Estimator (RSBE) using signed, prefix-sensitive NDKL to quantify remaining skew; and (iii) Confidence-Gated Logit Calibration (CGLC) that converts the residual signal into small and margin-focused logit corrections without retraining. On an occupation classification task, our approach reduces retriever-side skew (lowest NDKL among baselines for both dense and sparse retrievers) and achieves the lowest generator-side disparity (e.g., Risk Difference) while largely preserving utility. The same calibration can be tuned to alternative fairness criteria (e.g., Equal Opportunity) with minimal utility loss.</abstract>
    <identifier type="citekey">haque-etal-2026-fair</identifier>
    <identifier type="doi">10.18653/v1/2026.findings-acl.1358</identifier>
    <location>
        <url>https://aclanthology.org/2026.findings-acl.1358/</url>
    </location>
    <part>
        <date>2026-07</date>
        <extent unit="page">
            <start>27256</start>
            <end>27270</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Fair RAG: End-to-End Fairness Across Retrieval and Generation
%A Haque, Farsheed
%A Fu, Zhe
%A Aditya, Ramit
%A Xu, Depeng
%A Niu, Xi
%Y Liakata, Maria
%Y Moreira, Viviane P.
%Y Zhang, Jiajun
%Y Jurgens, David
%S Findings of the Association for Computational Linguistics: ACL 2026
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, California, United States
%@ 979-8-89176-395-1
%F haque-etal-2026-fair
%X Large Language Models (LLMs) used in Retrieval-Augmented Generation (RAG) can amplify demographic bias: retrievers may surface skewed context and generators can propagate that skew into decisions. Prior work typically treats fairness in retrieval or generation in isolation, leaving end-to-end fairness in RAG underexplored. We propose a post-hoc pipeline that jointly controls both stages: (i) a Fair Greedy Reranker (FGR) that builds prefix-balanced slates toward a target group mix; (ii) a Residual Slate Bias Estimator (RSBE) using signed, prefix-sensitive NDKL to quantify remaining skew; and (iii) Confidence-Gated Logit Calibration (CGLC) that converts the residual signal into small and margin-focused logit corrections without retraining. On an occupation classification task, our approach reduces retriever-side skew (lowest NDKL among baselines for both dense and sparse retrievers) and achieves the lowest generator-side disparity (e.g., Risk Difference) while largely preserving utility. The same calibration can be tuned to alternative fairness criteria (e.g., Equal Opportunity) with minimal utility loss.
%R 10.18653/v1/2026.findings-acl.1358
%U https://aclanthology.org/2026.findings-acl.1358/
%U https://doi.org/10.18653/v1/2026.findings-acl.1358
%P 27256-27270

Markdown (Informal)

[Fair RAG: End-to-End Fairness Across Retrieval and Generation](https://aclanthology.org/2026.findings-acl.1358/) (Haque et al., Findings 2026)

• Fair RAG: End-to-End Fairness Across Retrieval and Generation (Haque et al., Findings 2026)

ACL

• Farsheed Haque, Zhe Fu, Ramit Aditya, Depeng Xu, and Xi Niu. 2026. Fair RAG: End-to-End Fairness Across Retrieval and Generation. In Findings of the Association for Computational Linguistics: ACL 2026, pages 27256–27270, San Diego, California, United States. Association for Computational Linguistics.