The process of finding correspondences between concepts in different ontologies to enable interoperability.
Ontology alignment (also called ontology matching or ontology mapping) is the process of identifying correspondences between concepts, properties, and relationships in two or more different ontologies. When different organizations, systems, or domains use different ontologies to represent similar knowledge, alignment is necessary to enable semantic interoperability — allowing systems to understand and exchange information across ontological boundaries.
As enterprise AI deployments span multiple vendors, cloud providers, and business units — each with their own data models and ontologies — ontology alignment has become a critical engineering discipline. The Open Semantic Interchange (OSI) initiative, launched by Snowflake and partners in 2025, is attempting to create a common alignment framework for the industry. Organizations that master ontology alignment can integrate AI systems across their entire data estate.
Ontology alignment uses a combination of linguistic similarity (comparing concept names and descriptions), structural similarity (comparing the position of concepts in their respective hierarchies), and semantic similarity (comparing the meaning of concepts using embeddings). Modern alignment systems use ML models trained on known correspondences to predict alignments in new ontology pairs. The output is an alignment file specifying equivalences, subsumptions, and partial matches between concepts.
A hospital system merges with another hospital. The first uses SNOMED CT for clinical terminology; the second uses ICD-11. Ontology alignment maps 'Acute Myocardial Infarction' in SNOMED to 'ST elevation myocardial infarction' and 'Non-ST elevation myocardial infarction' in ICD-11, enabling the merged system's AI to correctly understand and analyze clinical data from both sources.