Ontology Embeddings

Ontology embeddings represent knowledge graphs and ontologies as vectors in a low-dimensional space, enabling efficient similarity computations and inference of missing relationships. Current research focuses on improving embedding methods, particularly addressing challenges like handling diverse relationship types (one-to-many, many-to-many), incorporating deductive closure and negative sampling strategies, and leveraging contextual information from surrounding concepts and axioms. These advancements are improving performance in tasks such as ontology completion, subsumption prediction, and claim detection, with applications in knowledge base construction, automated reasoning, and enhancing machine learning models with domain knowledge.