@inproceedings{shen-zhang-2026-preference,
title = "Preference-Aligned Data Synthesis for Fine-Grained Intent Detection",
author = "Shen, Sitong and
Zhang, Xiaotong",
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.163/",
pages = "3742--3758",
ISBN = "979-8-89176-459-0",
abstract = "Fine-grained intent detection becomes brittle as label spaces expand with many semantically adjacent intents. Although large language models (LLMs) can synthesize labeled utterances to reduce annotation cost, the resulting pseudo data are often ambiguous across intents and repetitive within intents. Existing synthesis pipelines mainly rely on prompt engineering and filtering, without updating generator parameters. We propose a preference-aligned synthesis framework that automatically mines supervision to optimize the generator. Each intent is represented by multi-centroid prototypes, and block-conditioned generation supports diversity-aware candidate construction. From these candidates, we build preferred blocks that are both prototype-consistent and diverse, and rejected blocks composed of confusable negatives from neighboring intents. These preference pairs are used to optimize the generator with Direct Preference Optimization (DPO), together with iterative refresh to mine harder confusions over time. Experiments show consistent gains for downstream intent classification in zero-shot and few-shot settings, while remaining competitive in fully supervised settings. We further observe improved generator reliability and strong transfer across multiple backbone LLM families."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="shen-zhang-2026-preference">
<titleInfo>
<title>Preference-Aligned Data Synthesis for Fine-Grained Intent Detection</title>
</titleInfo>
<name type="personal">
<namePart type="given">Sitong</namePart>
<namePart type="family">Shen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiaotong</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>Fine-grained intent detection becomes brittle as label spaces expand with many semantically adjacent intents. Although large language models (LLMs) can synthesize labeled utterances to reduce annotation cost, the resulting pseudo data are often ambiguous across intents and repetitive within intents. Existing synthesis pipelines mainly rely on prompt engineering and filtering, without updating generator parameters. We propose a preference-aligned synthesis framework that automatically mines supervision to optimize the generator. Each intent is represented by multi-centroid prototypes, and block-conditioned generation supports diversity-aware candidate construction. From these candidates, we build preferred blocks that are both prototype-consistent and diverse, and rejected blocks composed of confusable negatives from neighboring intents. These preference pairs are used to optimize the generator with Direct Preference Optimization (DPO), together with iterative refresh to mine harder confusions over time. Experiments show consistent gains for downstream intent classification in zero-shot and few-shot settings, while remaining competitive in fully supervised settings. We further observe improved generator reliability and strong transfer across multiple backbone LLM families.</abstract>
<identifier type="citekey">shen-zhang-2026-preference</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.163/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>3742</start>
<end>3758</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Preference-Aligned Data Synthesis for Fine-Grained Intent Detection
%A Shen, Sitong
%A Zhang, Xiaotong
%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 shen-zhang-2026-preference
%X Fine-grained intent detection becomes brittle as label spaces expand with many semantically adjacent intents. Although large language models (LLMs) can synthesize labeled utterances to reduce annotation cost, the resulting pseudo data are often ambiguous across intents and repetitive within intents. Existing synthesis pipelines mainly rely on prompt engineering and filtering, without updating generator parameters. We propose a preference-aligned synthesis framework that automatically mines supervision to optimize the generator. Each intent is represented by multi-centroid prototypes, and block-conditioned generation supports diversity-aware candidate construction. From these candidates, we build preferred blocks that are both prototype-consistent and diverse, and rejected blocks composed of confusable negatives from neighboring intents. These preference pairs are used to optimize the generator with Direct Preference Optimization (DPO), together with iterative refresh to mine harder confusions over time. Experiments show consistent gains for downstream intent classification in zero-shot and few-shot settings, while remaining competitive in fully supervised settings. We further observe improved generator reliability and strong transfer across multiple backbone LLM families.
%U https://aclanthology.org/2026.emnlp-main.163/
%P 3742-3758
Markdown (Informal)
[Preference-Aligned Data Synthesis for Fine-Grained Intent Detection](https://aclanthology.org/2026.emnlp-main.163/) (Shen & Zhang, EMNLP 2026)
ACL