@inproceedings{han-etal-2026-distilling,
title = "Distilling {LLM} Reasoning into Dense Encoders: Bridging the Accuracy-Efficiency Gap in Recommendation",
author = "Han, Donghee and
Roh, Daeyoung and
Kim, A Young and
Song, Hwanjun and
Yi, Mun Yong",
editor = "Liakata, Maria and
Moreira, Viviane P. and
Zhang, Jiajun and
Jurgens, David",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {ACL} 2026",
month = jul,
year = "2026",
address = "San Diego, California, United States",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-acl.1130/",
pages = "22513--22531",
ISBN = "979-8-89176-395-1",
abstract = "Large Language Models (LLMs) have shown remarkable potential in recommendation systems but suffer from prohibitive inference latency. Existing distillation approaches typically target Small Language Models (SLMs) or Conventional Recommendation Models (CRMs), face a critical trade-off between computational cost and semantic reasoning capacity. To bridge this accuracy-efficiency gap, we introduce Reasoning-to-Encoder Distillation (R2END), a framework that establishes a text encoder as the optimal student architecture for scalable recommendation. Unlike methods that mimic token generation, R2END compresses the teacher{'}s reasoning into a dense vector space via a semantic alignment objective, effectively capturing user-item dynamics. Extensive experiments on four datasets demonstrate that R2END not only outperforms state-of-the-art baselines but also achieves drastically reduced latency, offering a sweet spot for recommendation."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="han-etal-2026-distilling">
<titleInfo>
<title>Distilling LLM Reasoning into Dense Encoders: Bridging the Accuracy-Efficiency Gap in Recommendation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Donghee</namePart>
<namePart type="family">Han</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Daeyoung</namePart>
<namePart type="family">Roh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">A</namePart>
<namePart type="given">Young</namePart>
<namePart type="family">Kim</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hwanjun</namePart>
<namePart type="family">Song</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mun</namePart>
<namePart type="given">Yong</namePart>
<namePart type="family">Yi</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) have shown remarkable potential in recommendation systems but suffer from prohibitive inference latency. Existing distillation approaches typically target Small Language Models (SLMs) or Conventional Recommendation Models (CRMs), face a critical trade-off between computational cost and semantic reasoning capacity. To bridge this accuracy-efficiency gap, we introduce Reasoning-to-Encoder Distillation (R2END), a framework that establishes a text encoder as the optimal student architecture for scalable recommendation. Unlike methods that mimic token generation, R2END compresses the teacher’s reasoning into a dense vector space via a semantic alignment objective, effectively capturing user-item dynamics. Extensive experiments on four datasets demonstrate that R2END not only outperforms state-of-the-art baselines but also achieves drastically reduced latency, offering a sweet spot for recommendation.</abstract>
<identifier type="citekey">han-etal-2026-distilling</identifier>
<location>
<url>https://aclanthology.org/2026.findings-acl.1130/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>22513</start>
<end>22531</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Distilling LLM Reasoning into Dense Encoders: Bridging the Accuracy-Efficiency Gap in Recommendation
%A Han, Donghee
%A Roh, Daeyoung
%A Kim, A. Young
%A Song, Hwanjun
%A Yi, Mun Yong
%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 han-etal-2026-distilling
%X Large Language Models (LLMs) have shown remarkable potential in recommendation systems but suffer from prohibitive inference latency. Existing distillation approaches typically target Small Language Models (SLMs) or Conventional Recommendation Models (CRMs), face a critical trade-off between computational cost and semantic reasoning capacity. To bridge this accuracy-efficiency gap, we introduce Reasoning-to-Encoder Distillation (R2END), a framework that establishes a text encoder as the optimal student architecture for scalable recommendation. Unlike methods that mimic token generation, R2END compresses the teacher’s reasoning into a dense vector space via a semantic alignment objective, effectively capturing user-item dynamics. Extensive experiments on four datasets demonstrate that R2END not only outperforms state-of-the-art baselines but also achieves drastically reduced latency, offering a sweet spot for recommendation.
%U https://aclanthology.org/2026.findings-acl.1130/
%P 22513-22531
Markdown (Informal)
[Distilling LLM Reasoning into Dense Encoders: Bridging the Accuracy-Efficiency Gap in Recommendation](https://aclanthology.org/2026.findings-acl.1130/) (Han et al., Findings 2026)
ACL