@inproceedings{lyu-etal-2026-investigating,
title = "Investigating and Enhancing Concept Induction in {LLM}s: A Prototype-based Property-augmented Approach",
author = "Lyu, Mingyang and
Zhang, Bin and
Bao, Chenlong and
Tang, Jin-Tao and
Li, Shasha and
Wang, Ting",
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.225/",
pages = "5203--5219",
ISBN = "979-8-89176-459-0",
abstract = "Concept induction, which infers appropriate abstract categories from concrete instances, is both a core ontology engineering task and a testbed for LLM semantic abstraction. Although LLMs offer broad knowledge and reasoning that may ease expert-driven ontology construction, prior work shows that knowledge deficiencies still limit them in specialized domains, while systematic evaluation of their induction failures remains lacking. To address this gap, we propose a familiarity-controlled evaluation framework that uses knowledge probing to separate item familiarity from concept induction ability. The evaluation reveals a systematic tendency to induce overly general concepts: LLMs often produce semantically correct but overly broad labels, and direct property augmentation improves performance without resolving this granularity mismatch. We further propose PPACI, an ontology-prior-guided approach that builds concept prototypes by integrating concept definitions with member item properties. For each query group, PPACI uses item properties to retrieve local item-level and global group-level candidate concepts as induction anchors. Experiments across multiple LLMs show that PPACI reduces overly general predictions by 17{--}28 percentage points while maintaining competitive induction quality. Our code and data are publicly available at https://github.com/Orchideous77/PPACI."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="lyu-etal-2026-investigating">
<titleInfo>
<title>Investigating and Enhancing Concept Induction in LLMs: A Prototype-based Property-augmented Approach</title>
</titleInfo>
<name type="personal">
<namePart type="given">Mingyang</namePart>
<namePart type="family">Lyu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bin</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chenlong</namePart>
<namePart type="family">Bao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jin-Tao</namePart>
<namePart type="family">Tang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shasha</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ting</namePart>
<namePart type="family">Wang</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>Concept induction, which infers appropriate abstract categories from concrete instances, is both a core ontology engineering task and a testbed for LLM semantic abstraction. Although LLMs offer broad knowledge and reasoning that may ease expert-driven ontology construction, prior work shows that knowledge deficiencies still limit them in specialized domains, while systematic evaluation of their induction failures remains lacking. To address this gap, we propose a familiarity-controlled evaluation framework that uses knowledge probing to separate item familiarity from concept induction ability. The evaluation reveals a systematic tendency to induce overly general concepts: LLMs often produce semantically correct but overly broad labels, and direct property augmentation improves performance without resolving this granularity mismatch. We further propose PPACI, an ontology-prior-guided approach that builds concept prototypes by integrating concept definitions with member item properties. For each query group, PPACI uses item properties to retrieve local item-level and global group-level candidate concepts as induction anchors. Experiments across multiple LLMs show that PPACI reduces overly general predictions by 17–28 percentage points while maintaining competitive induction quality. Our code and data are publicly available at https://github.com/Orchideous77/PPACI.</abstract>
<identifier type="citekey">lyu-etal-2026-investigating</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.225/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>5203</start>
<end>5219</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Investigating and Enhancing Concept Induction in LLMs: A Prototype-based Property-augmented Approach
%A Lyu, Mingyang
%A Zhang, Bin
%A Bao, Chenlong
%A Tang, Jin-Tao
%A Li, Shasha
%A Wang, Ting
%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 lyu-etal-2026-investigating
%X Concept induction, which infers appropriate abstract categories from concrete instances, is both a core ontology engineering task and a testbed for LLM semantic abstraction. Although LLMs offer broad knowledge and reasoning that may ease expert-driven ontology construction, prior work shows that knowledge deficiencies still limit them in specialized domains, while systematic evaluation of their induction failures remains lacking. To address this gap, we propose a familiarity-controlled evaluation framework that uses knowledge probing to separate item familiarity from concept induction ability. The evaluation reveals a systematic tendency to induce overly general concepts: LLMs often produce semantically correct but overly broad labels, and direct property augmentation improves performance without resolving this granularity mismatch. We further propose PPACI, an ontology-prior-guided approach that builds concept prototypes by integrating concept definitions with member item properties. For each query group, PPACI uses item properties to retrieve local item-level and global group-level candidate concepts as induction anchors. Experiments across multiple LLMs show that PPACI reduces overly general predictions by 17–28 percentage points while maintaining competitive induction quality. Our code and data are publicly available at https://github.com/Orchideous77/PPACI.
%U https://aclanthology.org/2026.emnlp-main.225/
%P 5203-5219
Markdown (Informal)
[Investigating and Enhancing Concept Induction in LLMs: A Prototype-based Property-augmented Approach](https://aclanthology.org/2026.emnlp-main.225/) (Lyu et al., EMNLP 2026)
ACL