@inproceedings{zaera-etal-2025-efficient,
title = "Efficient Out-of-Scope Detection in Dialogue Systems via Uncertainty-Driven {LLM} Routing",
author = "Zaera, {\'A}lvaro and
Popa, Diana Nicoleta and
Sekulic, Ivan and
Rosso, Paolo",
editor = "Rehm, Georg and
Li, Yunyao",
booktitle = "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 6: Industry Track)",
month = jul,
year = "2025",
address = "Vienna, Austria",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.acl-industry.25/",
doi = "10.18653/v1/2025.acl-industry.25",
pages = "328--335",
ISBN = "979-8-89176-288-6",
abstract = "Out-of-scope (OOS) intent detection is a critical challenge in task-oriented dialogue systems (TODS), as it ensures robustness to unseen and ambiguous queries. In this work, we propose a novel but simple modular framework that combines uncertainty modeling with fine-tuned large language models (LLMs) for efficient and accurate OOS detection. The first step applies uncertainty estimation to the output of an in-scope intent detection classifier, which is currently deployed in a real-world TODS handling tens of thousands of user interactions daily. The second step then leverages an emerging LLM-based approach, where a fine-tuned LLM is triggered to make a final decision on instances with high uncertainty.Unlike prior approaches, our method effectively balances computational efficiency and performance, combining traditional approaches with LLMs and yielding state-of-the-art results on key OOS detection benchmarks, including real-world OOS data acquired from a deployed TODS."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zaera-etal-2025-efficient">
<titleInfo>
<title>Efficient Out-of-Scope Detection in Dialogue Systems via Uncertainty-Driven LLM Routing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Álvaro</namePart>
<namePart type="family">Zaera</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Diana</namePart>
<namePart type="given">Nicoleta</namePart>
<namePart type="family">Popa</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ivan</namePart>
<namePart type="family">Sekulic</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Paolo</namePart>
<namePart type="family">Rosso</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 6: Industry Track)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Georg</namePart>
<namePart type="family">Rehm</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yunyao</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Vienna, Austria</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-288-6</identifier>
</relatedItem>
<abstract>Out-of-scope (OOS) intent detection is a critical challenge in task-oriented dialogue systems (TODS), as it ensures robustness to unseen and ambiguous queries. In this work, we propose a novel but simple modular framework that combines uncertainty modeling with fine-tuned large language models (LLMs) for efficient and accurate OOS detection. The first step applies uncertainty estimation to the output of an in-scope intent detection classifier, which is currently deployed in a real-world TODS handling tens of thousands of user interactions daily. The second step then leverages an emerging LLM-based approach, where a fine-tuned LLM is triggered to make a final decision on instances with high uncertainty.Unlike prior approaches, our method effectively balances computational efficiency and performance, combining traditional approaches with LLMs and yielding state-of-the-art results on key OOS detection benchmarks, including real-world OOS data acquired from a deployed TODS.</abstract>
<identifier type="citekey">zaera-etal-2025-efficient</identifier>
<identifier type="doi">10.18653/v1/2025.acl-industry.25</identifier>
<location>
<url>https://aclanthology.org/2025.acl-industry.25/</url>
</location>
<part>
<date>2025-07</date>
<extent unit="page">
<start>328</start>
<end>335</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Efficient Out-of-Scope Detection in Dialogue Systems via Uncertainty-Driven LLM Routing
%A Zaera, Álvaro
%A Popa, Diana Nicoleta
%A Sekulic, Ivan
%A Rosso, Paolo
%Y Rehm, Georg
%Y Li, Yunyao
%S Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 6: Industry Track)
%D 2025
%8 July
%I Association for Computational Linguistics
%C Vienna, Austria
%@ 979-8-89176-288-6
%F zaera-etal-2025-efficient
%X Out-of-scope (OOS) intent detection is a critical challenge in task-oriented dialogue systems (TODS), as it ensures robustness to unseen and ambiguous queries. In this work, we propose a novel but simple modular framework that combines uncertainty modeling with fine-tuned large language models (LLMs) for efficient and accurate OOS detection. The first step applies uncertainty estimation to the output of an in-scope intent detection classifier, which is currently deployed in a real-world TODS handling tens of thousands of user interactions daily. The second step then leverages an emerging LLM-based approach, where a fine-tuned LLM is triggered to make a final decision on instances with high uncertainty.Unlike prior approaches, our method effectively balances computational efficiency and performance, combining traditional approaches with LLMs and yielding state-of-the-art results on key OOS detection benchmarks, including real-world OOS data acquired from a deployed TODS.
%R 10.18653/v1/2025.acl-industry.25
%U https://aclanthology.org/2025.acl-industry.25/
%U https://doi.org/10.18653/v1/2025.acl-industry.25
%P 328-335
Markdown (Informal)
[Efficient Out-of-Scope Detection in Dialogue Systems via Uncertainty-Driven LLM Routing](https://aclanthology.org/2025.acl-industry.25/) (Zaera et al., ACL 2025)
ACL