@inproceedings{lu-etal-2026-lore,
title = "{L}o{RE}: Enhancing Search Relevance with Progressive Chain-of-Thought and Preference Alignment",
author = "Lu, Chenji and
Chen, Zhuo and
Zhao, Hui and
Zeng, Zhiyuan and
Zhao, Gang and
Ren, Junjie and
Lihaoran and
Liu, Songyan and
Wang, Pengjie and
Yu, Chuan and
Xu, Jian and
Zheng, Bo",
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.1536/",
pages = "30754--30768",
ISBN = "979-8-89176-395-1",
abstract = "E-commerce search relevance is a critical component of retrieval systems. While Large Language Models (LLMs)-driven Chain-of-Thought (CoT) modeling has become the dominant paradigm and yielded significant gains, a critical gap remains: the absence of a systematic definition for comprehensive relevance reasoning, which leads to significant blind spots in current approaches. In this paper, we deconstruct the task into three core competencies: reasoning knowledge, multi-modal understanding, and rule awareness. Accordingly, we propose LoRE(Large Generative Model for Search Relevance), a novel two-stage training framework. We first employ an SFT phase to instill these capabilities via a progressive CoT synthesis pipeline, followed by a Reinforcement Learning(RL) phase, which serves as a regularizer, pruning redundant logic to achieve precise and robust adjudication. Extensive experiments validate LoRE, outperforming GPT-5 by 29.1{\%} in Macro-F1 and achieving a 27{\%} online gain, offering a vital reference for industrial domain-specific post-training."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="lu-etal-2026-lore">
<titleInfo>
<title>LoRE: Enhancing Search Relevance with Progressive Chain-of-Thought and Preference Alignment</title>
</titleInfo>
<name type="personal">
<namePart type="given">Chenji</namePart>
<namePart type="family">Lu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhuo</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hui</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhiyuan</namePart>
<namePart type="family">Zeng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gang</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Junjie</namePart>
<namePart type="family">Ren</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name>
<namePart>Lihaoran</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Songyan</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pengjie</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chuan</namePart>
<namePart type="family">Yu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jian</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bo</namePart>
<namePart type="family">Zheng</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>E-commerce search relevance is a critical component of retrieval systems. While Large Language Models (LLMs)-driven Chain-of-Thought (CoT) modeling has become the dominant paradigm and yielded significant gains, a critical gap remains: the absence of a systematic definition for comprehensive relevance reasoning, which leads to significant blind spots in current approaches. In this paper, we deconstruct the task into three core competencies: reasoning knowledge, multi-modal understanding, and rule awareness. Accordingly, we propose LoRE(Large Generative Model for Search Relevance), a novel two-stage training framework. We first employ an SFT phase to instill these capabilities via a progressive CoT synthesis pipeline, followed by a Reinforcement Learning(RL) phase, which serves as a regularizer, pruning redundant logic to achieve precise and robust adjudication. Extensive experiments validate LoRE, outperforming GPT-5 by 29.1% in Macro-F1 and achieving a 27% online gain, offering a vital reference for industrial domain-specific post-training.</abstract>
<identifier type="citekey">lu-etal-2026-lore</identifier>
<location>
<url>https://aclanthology.org/2026.findings-acl.1536/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>30754</start>
<end>30768</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T LoRE: Enhancing Search Relevance with Progressive Chain-of-Thought and Preference Alignment
%A Lu, Chenji
%A Chen, Zhuo
%A Zhao, Hui
%A Zeng, Zhiyuan
%A Zhao, Gang
%A Ren, Junjie
%A Liu, Songyan
%A Wang, Pengjie
%A Yu, Chuan
%A Xu, Jian
%A Zheng, Bo
%Y Liakata, Maria
%Y Moreira, Viviane P.
%Y Zhang, Jiajun
%Y Jurgens, David
%A Lihaoran
%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 lu-etal-2026-lore
%X E-commerce search relevance is a critical component of retrieval systems. While Large Language Models (LLMs)-driven Chain-of-Thought (CoT) modeling has become the dominant paradigm and yielded significant gains, a critical gap remains: the absence of a systematic definition for comprehensive relevance reasoning, which leads to significant blind spots in current approaches. In this paper, we deconstruct the task into three core competencies: reasoning knowledge, multi-modal understanding, and rule awareness. Accordingly, we propose LoRE(Large Generative Model for Search Relevance), a novel two-stage training framework. We first employ an SFT phase to instill these capabilities via a progressive CoT synthesis pipeline, followed by a Reinforcement Learning(RL) phase, which serves as a regularizer, pruning redundant logic to achieve precise and robust adjudication. Extensive experiments validate LoRE, outperforming GPT-5 by 29.1% in Macro-F1 and achieving a 27% online gain, offering a vital reference for industrial domain-specific post-training.
%U https://aclanthology.org/2026.findings-acl.1536/
%P 30754-30768
Markdown (Informal)
[LoRE: Enhancing Search Relevance with Progressive Chain-of-Thought and Preference Alignment](https://aclanthology.org/2026.findings-acl.1536/) (Lu et al., Findings 2026)
ACL
- Chenji Lu, Zhuo Chen, Hui Zhao, Zhiyuan Zeng, Gang Zhao, Junjie Ren, Lihaoran, Songyan Liu, Pengjie Wang, Chuan Yu, Jian Xu, and Bo Zheng. 2026. LoRE: Enhancing Search Relevance with Progressive Chain-of-Thought and Preference Alignment. In Findings of the Association for Computational Linguistics: ACL 2026, pages 30754–30768, San Diego, California, United States. Association for Computational Linguistics.