Ontology alignment is a large area of research within the semantic web. It involves the generation of mappings between concepts across two ontologies, which are commonly defined as the source and the target. These mappings indicate the existence of related concepts across the two ontologies facilitating cross-ontology reasoning and interoperability. Recent works have demonstrated the viability of alignments as an ontology evaluation measure [1,2]. If an equivalent representation of a given concept exists in both an established, community verified ontology and a new ontology, this alignment provides evidence that the new ontology follows a compatible view of the domain. This use of ontology alignments, whist useful, requires significant oversight from the ontology engineer and domain expert. With an increasing number of works exploring the possibility of automatically, or semi-automatically, generating ontologies [3], the volume of human overhead will significantly increase. We hypothesise that the amount of these alignments that would need manual verification could be significantly reduced if it were possible to estimate the expected number of alignments between two ontologies prior to the generation of these alignments. A significant difference in the number of expected and actual alignments could indicate a difference in representation between the ontologies that would then prompt human intervention. To test this hypothesis, we estimate the overlap between two ontologies via a suite of graph similarity measures, we then use this overlap to then estimate the number of alignments between the two ontologies. This estimate is then compared against a ground truth set of reference alignments defined as part of the Ontology Alignment Evaluation Initiative.
[1] Wilf Morlidge, Elliott Watkiss-Leek, George Hannah, Harry Rostron, Andrew Ng, Ewan Johnson, Andrew Mitchell, Terry R. Payne, Valentina Tamma, and Jacopo de Berardinis. The animl ontology: Enabling semantic interoperability for large-scale experimental data in interconnected scientific labs. In Proceedings of the 38th International Conference on Advanced Information Systems Engineering (CAiSE), Verona, Italy, June 2026.
[2] George Hannah, Terry R Payne, Valentina Tamma, Andrew Mitchell, Ellen Piercy, and Boris Konev. Towards a methodology for the semi-automatic generation of scientific knowledge graphs from xml documents. In In: The 18th International Workshop on Ontology Matching, 2023-11-7 - 2023-11-7, Athens, Greece., 2023.
[3] George Hannah, Jacopo de Berardinis, Terry R Payne, Valentina Tamma, Andrew Mitchell, Ellen Piercy, Ewan Johnson, Andrew Ng, Harry Rostron, and Boris Konev. Relrae: Llm-based relationship extraction, labelling, refinement, and evaluation. arXiv preprint arXiv:2507.03829, 2025.