Universal Framework for Integrating New Concepts Into Ontologies Across Multiple Domains
Year: 2026
Type: conference-paper
Abstract: Ontologies must be regularly expanded through the incorporation of new concepts and relationships due to the ongoing evolution of topic areas. Conventional ontological evolution techniques are time-consuming, prone to error, and limited in scalability because they mostly rely on the manual labor of specialists. This article proposes a universal framework for integrating new concepts into the ontology of various subject areas based on the use of language models. The framework implements a three-step process that includes semantic candidate generation, contextual validation, and formal ontological placement of concepts. Language models are used to analyze textual definitions, calculate semantic similarities, and identify hierarchical relationships between concepts. Experimental evaluation was carried out on medical (SNOMED CT), industrial and educational ontologies. The experimental results demonstrate an increase in the accuracy of concept placement and significant reduction in the amount of expert corrections compared to traditional rule-based approaches. The proposed framework has versatility and can be used to support the evolution of ontologies in various applied fields.
Institution: Q University
Language: English
Topic: Semantic Web and Ontologies
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7168260599