@inproceedings{fan-etal-2026-rank4gen,
title = "{R}ank4{G}en: {RAG}-Preference-Aligned Document Set Selection and Ranking",
author = "Fan, Yongqi and
Chu, Yuxiang and
Xia, Zhentao and
Chen, Xiaoyang and
Liu, Jie and
Liang, Haijin and
Ma, Jin and
He, Ben and
Sun, Yingfei and
Zhai, Jie and
Ye, Dezhi and
Ruan, Tong",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.942/",
pages = "19847--19875",
ISBN = "979-8-89176-462-0",
abstract = "In the RAG paradigm, document ranking determines the evidence available to downstream generators. Through controlled analysis, we identify two phenomena underexplored by existing rankers: (i) downstream response quality depends not only on relevance but also on the composition and ordering of selected documents, and (ii) such preferences differ systematically across generators. However, existing rankers are trained purely on query{--}document relevance, leaving both phenomena unmodeled. To close this gap, we construct PRISM, a bilingual preference-aligned dataset built through a four-stage pipeline that compresses the combinatorial subset-and-ordering space by roughly four orders of magnitude and produces response-quality preference supervision conditioned on seven downstream generators. On a 13k-query subset of PRISM, we train Rank4Gen, a generator-aware ranker that performs joint document set selection and ordering. Experiments on five challenging RAG benchmarks show that Rank4Gen improves downstream QA quality on most evaluated generators, with per-generator F1 gains of up to +2.08 over the strongest set-selection baseline. Code is available at https://github.com/JOHNNY-fans/Rank4Gen."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="fan-etal-2026-rank4gen">
<titleInfo>
<title>Rank4Gen: RAG-Preference-Aligned Document Set Selection and Ranking</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yongqi</namePart>
<namePart type="family">Fan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yuxiang</namePart>
<namePart type="family">Chu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhentao</namePart>
<namePart type="family">Xia</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiaoyang</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jie</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haijin</namePart>
<namePart type="family">Liang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jin</namePart>
<namePart type="family">Ma</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ben</namePart>
<namePart type="family">He</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yingfei</namePart>
<namePart type="family">Sun</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jie</namePart>
<namePart type="family">Zhai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dezhi</namePart>
<namePart type="family">Ye</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tong</namePart>
<namePart type="family">Ruan</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>Findings of the Association for Computational Linguistics: EMNLP 2026</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">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>In the RAG paradigm, document ranking determines the evidence available to downstream generators. Through controlled analysis, we identify two phenomena underexplored by existing rankers: (i) downstream response quality depends not only on relevance but also on the composition and ordering of selected documents, and (ii) such preferences differ systematically across generators. However, existing rankers are trained purely on query–document relevance, leaving both phenomena unmodeled. To close this gap, we construct PRISM, a bilingual preference-aligned dataset built through a four-stage pipeline that compresses the combinatorial subset-and-ordering space by roughly four orders of magnitude and produces response-quality preference supervision conditioned on seven downstream generators. On a 13k-query subset of PRISM, we train Rank4Gen, a generator-aware ranker that performs joint document set selection and ordering. Experiments on five challenging RAG benchmarks show that Rank4Gen improves downstream QA quality on most evaluated generators, with per-generator F1 gains of up to +2.08 over the strongest set-selection baseline. Code is available at https://github.com/JOHNNY-fans/Rank4Gen.</abstract>
<identifier type="citekey">fan-etal-2026-rank4gen</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.942/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>19847</start>
<end>19875</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Rank4Gen: RAG-Preference-Aligned Document Set Selection and Ranking
%A Fan, Yongqi
%A Chu, Yuxiang
%A Xia, Zhentao
%A Chen, Xiaoyang
%A Liu, Jie
%A Liang, Haijin
%A Ma, Jin
%A He, Ben
%A Sun, Yingfei
%A Zhai, Jie
%A Ye, Dezhi
%A Ruan, Tong
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F fan-etal-2026-rank4gen
%X In the RAG paradigm, document ranking determines the evidence available to downstream generators. Through controlled analysis, we identify two phenomena underexplored by existing rankers: (i) downstream response quality depends not only on relevance but also on the composition and ordering of selected documents, and (ii) such preferences differ systematically across generators. However, existing rankers are trained purely on query–document relevance, leaving both phenomena unmodeled. To close this gap, we construct PRISM, a bilingual preference-aligned dataset built through a four-stage pipeline that compresses the combinatorial subset-and-ordering space by roughly four orders of magnitude and produces response-quality preference supervision conditioned on seven downstream generators. On a 13k-query subset of PRISM, we train Rank4Gen, a generator-aware ranker that performs joint document set selection and ordering. Experiments on five challenging RAG benchmarks show that Rank4Gen improves downstream QA quality on most evaluated generators, with per-generator F1 gains of up to +2.08 over the strongest set-selection baseline. Code is available at https://github.com/JOHNNY-fans/Rank4Gen.
%U https://aclanthology.org/2026.findings-emnlp.942/
%P 19847-19875
Markdown (Informal)
[Rank4Gen: RAG-Preference-Aligned Document Set Selection and Ranking](https://aclanthology.org/2026.findings-emnlp.942/) (Fan et al., Findings 2026)
ACL
- Yongqi Fan, Yuxiang Chu, Zhentao Xia, Xiaoyang Chen, Jie Liu, Haijin Liang, Jin Ma, Ben He, Yingfei Sun, Jie Zhai, Dezhi Ye, and Tong Ruan. 2026. Rank4Gen: RAG-Preference-Aligned Document Set Selection and Ranking. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 19847–19875, Budapest, Hungary. Association for Computational Linguistics.