@inproceedings{lin-etal-2026-taxoagent,
title = "{T}axo{A}gent: Utility-Aligned Policy Optimization for Hierarchical Semantic Classification",
author = "Lin, Zhiming and
Zhong, Yihao and
Hou, Liwei and
Hu, Saisai and
He, ShaoYan and
Rongru, Chen and
Yang, Xuan and
Liu, Siyue",
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.1006/",
pages = "23149--23171",
ISBN = "979-8-89176-459-0",
abstract = "Hierarchical semantic classification is essential for taxonomy-driven applications, yet existing classifiers often optimize flat label likelihoods and therefore weakly control parent-child validity, semantic error severity, coarse-to-fine coherence, and out-of-scope rejection. This paper aims to align LLM-based classifiers with the full structured utility of hierarchical prediction rather than treating taxonomy labels as independent targets. We propose TaxoAgent, a utility-aligned policy optimization framework that formulates taxonomy prediction as a verified parent-to-child decision trajectory. Given an input, TaxoAgent first predicts coarse parent labels, expands them into fine-grained child decisions, and verifies the resulting trajectory with a composite utility that jointly measures outcome correctness, hierarchy consistency, semantic proximity, and trajectory coherence; group-relative policy optimization then reinforces high-utility trajectories over low-utility alternatives. Across IOS-APP-ZH, Google-APP-EN, RCV1, WOS, and CLINC150, TaxoAgent consistently improves Micro-F1 and Macro-F1, reduces hierarchy inconsistency, strengthens long-tail classification, and improves OOS-F1 by up to 3.68 points over the strongest prior baseline. These results show that treating taxonomies as structured decision spaces offers a practical and extensible path toward more reliable, adaptable, and utility-aware semantic classifiers."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="lin-etal-2026-taxoagent">
<titleInfo>
<title>TaxoAgent: Utility-Aligned Policy Optimization for Hierarchical Semantic Classification</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zhiming</namePart>
<namePart type="family">Lin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yihao</namePart>
<namePart type="family">Zhong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Liwei</namePart>
<namePart type="family">Hou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Saisai</namePart>
<namePart type="family">Hu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">ShaoYan</namePart>
<namePart type="family">He</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chen</namePart>
<namePart type="family">Rongru</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xuan</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Siyue</namePart>
<namePart type="family">Liu</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>Hierarchical semantic classification is essential for taxonomy-driven applications, yet existing classifiers often optimize flat label likelihoods and therefore weakly control parent-child validity, semantic error severity, coarse-to-fine coherence, and out-of-scope rejection. This paper aims to align LLM-based classifiers with the full structured utility of hierarchical prediction rather than treating taxonomy labels as independent targets. We propose TaxoAgent, a utility-aligned policy optimization framework that formulates taxonomy prediction as a verified parent-to-child decision trajectory. Given an input, TaxoAgent first predicts coarse parent labels, expands them into fine-grained child decisions, and verifies the resulting trajectory with a composite utility that jointly measures outcome correctness, hierarchy consistency, semantic proximity, and trajectory coherence; group-relative policy optimization then reinforces high-utility trajectories over low-utility alternatives. Across IOS-APP-ZH, Google-APP-EN, RCV1, WOS, and CLINC150, TaxoAgent consistently improves Micro-F1 and Macro-F1, reduces hierarchy inconsistency, strengthens long-tail classification, and improves OOS-F1 by up to 3.68 points over the strongest prior baseline. These results show that treating taxonomies as structured decision spaces offers a practical and extensible path toward more reliable, adaptable, and utility-aware semantic classifiers.</abstract>
<identifier type="citekey">lin-etal-2026-taxoagent</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1006/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>23149</start>
<end>23171</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T TaxoAgent: Utility-Aligned Policy Optimization for Hierarchical Semantic Classification
%A Lin, Zhiming
%A Zhong, Yihao
%A Hou, Liwei
%A Hu, Saisai
%A He, ShaoYan
%A Rongru, Chen
%A Yang, Xuan
%A Liu, Siyue
%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 lin-etal-2026-taxoagent
%X Hierarchical semantic classification is essential for taxonomy-driven applications, yet existing classifiers often optimize flat label likelihoods and therefore weakly control parent-child validity, semantic error severity, coarse-to-fine coherence, and out-of-scope rejection. This paper aims to align LLM-based classifiers with the full structured utility of hierarchical prediction rather than treating taxonomy labels as independent targets. We propose TaxoAgent, a utility-aligned policy optimization framework that formulates taxonomy prediction as a verified parent-to-child decision trajectory. Given an input, TaxoAgent first predicts coarse parent labels, expands them into fine-grained child decisions, and verifies the resulting trajectory with a composite utility that jointly measures outcome correctness, hierarchy consistency, semantic proximity, and trajectory coherence; group-relative policy optimization then reinforces high-utility trajectories over low-utility alternatives. Across IOS-APP-ZH, Google-APP-EN, RCV1, WOS, and CLINC150, TaxoAgent consistently improves Micro-F1 and Macro-F1, reduces hierarchy inconsistency, strengthens long-tail classification, and improves OOS-F1 by up to 3.68 points over the strongest prior baseline. These results show that treating taxonomies as structured decision spaces offers a practical and extensible path toward more reliable, adaptable, and utility-aware semantic classifiers.
%U https://aclanthology.org/2026.emnlp-main.1006/
%P 23149-23171
Markdown (Informal)
[TaxoAgent: Utility-Aligned Policy Optimization for Hierarchical Semantic Classification](https://aclanthology.org/2026.emnlp-main.1006/) (Lin et al., EMNLP 2026)
ACL
- Zhiming Lin, Yihao Zhong, Liwei Hou, Saisai Hu, ShaoYan He, Chen Rongru, Xuan Yang, and Siyue Liu. 2026. TaxoAgent: Utility-Aligned Policy Optimization for Hierarchical Semantic Classification. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 23149–23171, Budapest, Hungary. Association for Computational Linguistics.