@inproceedings{sun-sang-2026-cost,
title = "Cost-Efficient Intent Classification via Semantic-Candidate Adaptation and Selective Capability Routing",
author = "Sun, Jiawei and
Sang, Yongsheng",
editor = "Atanasova, Pepa and
Haffari, Gholamreza and
Wang, Minghan",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: Industry Track",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-industry.257/",
pages = "4338--4348",
ISBN = "979-8-89176-460-6",
abstract = "Intent classification deployments trade accuracy against retraining effort and inference cost. We study frozen generative reranking and budgeted large-model escalation, introducing Semantic-Candidate Adaptation (SCA): BGE top-5 retrieval followed by joint reranking with one representative training example per candidate. On CLINC150, aligned examples raise Qwen2.5-3B-Instruct from 68.18{\%} label-only accuracy to 76.13{\%} and Gemma-2-2B-it from 66.02{\%} to 76.15{\%}, without LLM parameter updates. CLINC150 wrong-example controls support the role of correct alignment, while BANKING77 and HWU64 provide transfer evidence. Broader comparisons clarify the deployment regimes: direct centroid retrieval reaches 80.71{\%}, while a task-trained MPNet Energy baseline reaches 90.05{\%}; always-on SCA is therefore not a universal replacement for direct or trained classifiers. A validation-selected low-margin SCA fallback provides only a marginal gain at an 8.62{\%} realized test call rate. For real Qwen3.7-Max escalation on CLINC150, routing the lowest BGE top-1 retrieval scores reaches 82.11{--}82.89{\%} accuracy at a 10{\%} budget and 87.33{--}88.07{\%} at 20{\%}, with OOS F1 of 70.00{--}70.13{\%} and 87.47{--}87.70{\%}, respectively. Learned and token-confidence routing remain alternatives, but no policy dominates every evaluated point."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="sun-sang-2026-cost">
<titleInfo>
<title>Cost-Efficient Intent Classification via Semantic-Candidate Adaptation and Selective Capability Routing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jiawei</namePart>
<namePart type="family">Sun</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yongsheng</namePart>
<namePart type="family">Sang</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: Industry Track</title>
</titleInfo>
<name type="personal">
<namePart type="given">Pepa</namePart>
<namePart type="family">Atanasova</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gholamreza</namePart>
<namePart type="family">Haffari</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Minghan</namePart>
<namePart type="family">Wang</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">979-8-89176-460-6</identifier>
</relatedItem>
<abstract>Intent classification deployments trade accuracy against retraining effort and inference cost. We study frozen generative reranking and budgeted large-model escalation, introducing Semantic-Candidate Adaptation (SCA): BGE top-5 retrieval followed by joint reranking with one representative training example per candidate. On CLINC150, aligned examples raise Qwen2.5-3B-Instruct from 68.18% label-only accuracy to 76.13% and Gemma-2-2B-it from 66.02% to 76.15%, without LLM parameter updates. CLINC150 wrong-example controls support the role of correct alignment, while BANKING77 and HWU64 provide transfer evidence. Broader comparisons clarify the deployment regimes: direct centroid retrieval reaches 80.71%, while a task-trained MPNet Energy baseline reaches 90.05%; always-on SCA is therefore not a universal replacement for direct or trained classifiers. A validation-selected low-margin SCA fallback provides only a marginal gain at an 8.62% realized test call rate. For real Qwen3.7-Max escalation on CLINC150, routing the lowest BGE top-1 retrieval scores reaches 82.11–82.89% accuracy at a 10% budget and 87.33–88.07% at 20%, with OOS F1 of 70.00–70.13% and 87.47–87.70%, respectively. Learned and token-confidence routing remain alternatives, but no policy dominates every evaluated point.</abstract>
<identifier type="citekey">sun-sang-2026-cost</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-industry.257/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>4338</start>
<end>4348</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Cost-Efficient Intent Classification via Semantic-Candidate Adaptation and Selective Capability Routing
%A Sun, Jiawei
%A Sang, Yongsheng
%Y Atanasova, Pepa
%Y Haffari, Gholamreza
%Y Wang, Minghan
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: Industry Track
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-460-6
%F sun-sang-2026-cost
%X Intent classification deployments trade accuracy against retraining effort and inference cost. We study frozen generative reranking and budgeted large-model escalation, introducing Semantic-Candidate Adaptation (SCA): BGE top-5 retrieval followed by joint reranking with one representative training example per candidate. On CLINC150, aligned examples raise Qwen2.5-3B-Instruct from 68.18% label-only accuracy to 76.13% and Gemma-2-2B-it from 66.02% to 76.15%, without LLM parameter updates. CLINC150 wrong-example controls support the role of correct alignment, while BANKING77 and HWU64 provide transfer evidence. Broader comparisons clarify the deployment regimes: direct centroid retrieval reaches 80.71%, while a task-trained MPNet Energy baseline reaches 90.05%; always-on SCA is therefore not a universal replacement for direct or trained classifiers. A validation-selected low-margin SCA fallback provides only a marginal gain at an 8.62% realized test call rate. For real Qwen3.7-Max escalation on CLINC150, routing the lowest BGE top-1 retrieval scores reaches 82.11–82.89% accuracy at a 10% budget and 87.33–88.07% at 20%, with OOS F1 of 70.00–70.13% and 87.47–87.70%, respectively. Learned and token-confidence routing remain alternatives, but no policy dominates every evaluated point.
%U https://aclanthology.org/2026.emnlp-industry.257/
%P 4338-4348
Markdown (Informal)
[Cost-Efficient Intent Classification via Semantic-Candidate Adaptation and Selective Capability Routing](https://aclanthology.org/2026.emnlp-industry.257/) (Sun & Sang, EMNLP 2026)
ACL