@inproceedings{song-etal-2026-lq,
title = "{LQ}-Rewriter: Generative Reward Model-Guided Lazy Query Rewriting for Role-Playing Dialogue",
author = "Song, Wei and
Huang, Zheng and
Zhu, Renyu and
Wu, Runze and
Huang, Zhenya and
Mao, Yihuan and
Zhang, Rongsheng and
Lv, Tangjie and
Zhang, Kai",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.391/",
pages = "8977--8994",
ISBN = "979-8-89176-459-0",
abstract = "With the rapid development of large language models (LLMs), role-playing dialogue systems have become increasingly popular for immersive and personalized human{--}AI interaction. However, in real-world role-playing scenarios, users often provide lazy queries{---}brief and ambiguous responses such as ``oh,'' ``hmm,'' or ``?''{---}which lack explicit intent and emotional cues, making it difficult for role-playing LLMs to generate engaging replies. Existing query rewriting methods mainly target functional tasks such as information retrieval and are not tailored to role-playing dialogue, where rewrite quality depends on role-playing-specific factors such as character consistency and personalized user preferences. To address this, we first propose a listwise Generative Reward Model (GRM) trained from real user preference data collected through online regeneration behavior, which provides fine-grained process and final preference signals for training a Lazy Query Rewriter. Using rejection fine-tuning and reinforcement learning, the LQ-Rewriter learns to transform underspecified user inputs into richer expressions while preserving user intent and dialogue context. Experiments on real-world dialogue logs from a mainstream Chinese AI-powered romantic role-playing platform show that our GRM substantially outperforms general-purpose LLM judges. Moreover, our LQ-Rewriter achieves the highest GRM-based and human preference win rates among all rewriting baselines, and brings tangible improvements in online deployment."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="song-etal-2026-lq">
<titleInfo>
<title>LQ-Rewriter: Generative Reward Model-Guided Lazy Query Rewriting for Role-Playing Dialogue</title>
</titleInfo>
<name type="personal">
<namePart type="given">Wei</namePart>
<namePart type="family">Song</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zheng</namePart>
<namePart type="family">Huang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Renyu</namePart>
<namePart type="family">Zhu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Runze</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhenya</namePart>
<namePart type="family">Huang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yihuan</namePart>
<namePart type="family">Mao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rongsheng</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tangjie</namePart>
<namePart type="family">Lv</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kai</namePart>
<namePart type="family">Zhang</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>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</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">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>With the rapid development of large language models (LLMs), role-playing dialogue systems have become increasingly popular for immersive and personalized human–AI interaction. However, in real-world role-playing scenarios, users often provide lazy queries—brief and ambiguous responses such as “oh,” “hmm,” or “?”—which lack explicit intent and emotional cues, making it difficult for role-playing LLMs to generate engaging replies. Existing query rewriting methods mainly target functional tasks such as information retrieval and are not tailored to role-playing dialogue, where rewrite quality depends on role-playing-specific factors such as character consistency and personalized user preferences. To address this, we first propose a listwise Generative Reward Model (GRM) trained from real user preference data collected through online regeneration behavior, which provides fine-grained process and final preference signals for training a Lazy Query Rewriter. Using rejection fine-tuning and reinforcement learning, the LQ-Rewriter learns to transform underspecified user inputs into richer expressions while preserving user intent and dialogue context. Experiments on real-world dialogue logs from a mainstream Chinese AI-powered romantic role-playing platform show that our GRM substantially outperforms general-purpose LLM judges. Moreover, our LQ-Rewriter achieves the highest GRM-based and human preference win rates among all rewriting baselines, and brings tangible improvements in online deployment.</abstract>
<identifier type="citekey">song-etal-2026-lq</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.391/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>8977</start>
<end>8994</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T LQ-Rewriter: Generative Reward Model-Guided Lazy Query Rewriting for Role-Playing Dialogue
%A Song, Wei
%A Huang, Zheng
%A Zhu, Renyu
%A Wu, Runze
%A Huang, Zhenya
%A Mao, Yihuan
%A Zhang, Rongsheng
%A Lv, Tangjie
%A Zhang, Kai
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F song-etal-2026-lq
%X With the rapid development of large language models (LLMs), role-playing dialogue systems have become increasingly popular for immersive and personalized human–AI interaction. However, in real-world role-playing scenarios, users often provide lazy queries—brief and ambiguous responses such as “oh,” “hmm,” or “?”—which lack explicit intent and emotional cues, making it difficult for role-playing LLMs to generate engaging replies. Existing query rewriting methods mainly target functional tasks such as information retrieval and are not tailored to role-playing dialogue, where rewrite quality depends on role-playing-specific factors such as character consistency and personalized user preferences. To address this, we first propose a listwise Generative Reward Model (GRM) trained from real user preference data collected through online regeneration behavior, which provides fine-grained process and final preference signals for training a Lazy Query Rewriter. Using rejection fine-tuning and reinforcement learning, the LQ-Rewriter learns to transform underspecified user inputs into richer expressions while preserving user intent and dialogue context. Experiments on real-world dialogue logs from a mainstream Chinese AI-powered romantic role-playing platform show that our GRM substantially outperforms general-purpose LLM judges. Moreover, our LQ-Rewriter achieves the highest GRM-based and human preference win rates among all rewriting baselines, and brings tangible improvements in online deployment.
%U https://aclanthology.org/2026.emnlp-main.391/
%P 8977-8994
Markdown (Informal)
[LQ-Rewriter: Generative Reward Model-Guided Lazy Query Rewriting for Role-Playing Dialogue](https://aclanthology.org/2026.emnlp-main.391/) (Song et al., EMNLP 2026)
ACL
- Wei Song, Zheng Huang, Renyu Zhu, Runze Wu, Zhenya Huang, Yihuan Mao, Rongsheng Zhang, Tangjie Lv, and Kai Zhang. 2026. LQ-Rewriter: Generative Reward Model-Guided Lazy Query Rewriting for Role-Playing Dialogue. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 8977–8994, Budapest, Hungary. Association for Computational Linguistics.