Cross-Domain Classification of Domain Entities into Top-Level Ontology Concepts Using BERT: A Study Case on the BFO Domain Ontologies

Alcides Lopes, Joel Carbonera, Nicolau Santos, Fabricio Rodrigues, Luan Garcia, Mara Abel

2024

Abstract

Classifying domain entities into top-level ontology concepts using informal definitions remains an active research area with several open questions. One of these questions pertains to the quality of proposed pipelines employing language models for classifying informal definitions when training and testing samples are from different knowledge domains. This can introduce challenges due to varying vocabularies across domains or the potential for an entity to belong to different top-level concepts based on its domain. In this study, we present a study case where terms and informal definitions are extracted from 81 domain ontologies organized into 12 knowledge domains. We investigate the performance of a pipeline that utilizes the BERT language model for classifying domain entities into top-level concepts within a cross-domain classification scenario. Additionally, we explore various pipeline setups for input, preprocessing, and training steps. Our optimal classifier setup employs an unbalanced training methodology, no text preprocessing, and the concatenation of terms and informal definitions as input. Furthermore, we demonstrate that BERT yields promising results in classifying domain entities into top-level concepts within a cross-domain classification scenario.

Paper Citation

in Harvard Style

Lopes A., Carbonera J., Santos N., Rodrigues F., Garcia L. and Abel M. (2024). Cross-Domain Classification of Domain Entities into Top-Level Ontology Concepts Using BERT: A Study Case on the BFO Domain Ontologies. In Proceedings of the 26th International Conference on Enterprise Information Systems - Volume 2: ICEIS; ISBN 978-989-758-692-7, SciTePress, pages 141-148. DOI: 10.5220/0012557600003690

in Bibtex Style

@conference{iceis24,
author={Alcides Lopes and Joel Carbonera and Nicolau Santos and Fabricio Rodrigues and Luan Garcia and Mara Abel},
title={Cross-Domain Classification of Domain Entities into Top-Level Ontology Concepts Using BERT: A Study Case on the BFO Domain Ontologies},
booktitle={Proceedings of the 26th International Conference on Enterprise Information Systems - Volume 2: ICEIS},
year={2024},
pages={141-148},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012557600003690},
isbn={978-989-758-692-7},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 26th International Conference on Enterprise Information Systems - Volume 2: ICEIS
TI - Cross-Domain Classification of Domain Entities into Top-Level Ontology Concepts Using BERT: A Study Case on the BFO Domain Ontologies
SN - 978-989-758-692-7
AU - Lopes A.
AU - Carbonera J.
AU - Santos N.
AU - Rodrigues F.
AU - Garcia L.
AU - Abel M.
PY - 2024
SP - 141
EP - 148
DO - 10.5220/0012557600003690
PB - SciTePress