Year: 2026
Type: review
Abstract: Ontologies provide precise semantics and reliable reasoning capabilities and are widely used in real-world domains such as healthcare and finance. However, traditional ontology construction and reasoning face several limitations: ontology construction typically requires substantial human effort, while standard reasoning approaches struggle with incomplete or uncertain information. To address these challenges, the use of large language models (LLMs) in ontology construction and reasoning has gained significant attention. Despite a growing body of research, the field lacks a systematic review, which hinders researchers from obtaining a comprehensive understanding of recent advances. To bridge this gap, we survey recent work, primarily since 2021, and categorize the current research landscape into three areas: Ontology Learning, Ontology Matching, and Ontology-aware Complex Reasoning. We examine how ontologies are integrated with LLMs for knowledge extraction, interaction, and reasoning through approaches such as prompting, fine-tuning, and retrieval-augmented generation (RAG). Finally, we discuss current challenges and outline promising directions for future research.
Authors: Yang, Hui, Chen, Jiaoyan, Ma, Yue, Schmidt, Renate
Institutions: University of Manchester, Université Paris-Saclay
Language: English
Topic: Semantic Web and Ontologies
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7203845197
