Ontology

In AI and information science, an ontology is a formal, explicit specification of a shared conceptualisation. It defines the concepts that exist within a domain, the properties of those concepts, and the relationships between them. Unlike a simple glossary or vocabulary, an ontology captures the structure and logic of a knowledge domain. For example, a medical ontology might define that 'paracetamol' is a type of 'analgesic', which is a type of 'medication'. It would also capture relationships like 'treats' (paracetamol treats headache), 'interacts with' (paracetamol interacts with warfarin), and properties (dosage, route of administration). Ontologies range from lightweight taxonomies (simple hierarchical classifications) to heavyweight formal ontologies with axioms and logical rules. The appropriate level of formality depends on the use case — search and navigation benefit from lightweight ontologies, while automated reasoning requires more formal representations.

Ontologies provide the structured knowledge framework that makes AI systems more accurate and consistent. In enterprise contexts, ontologies standardise terminology across departments, enable interoperability between systems, and provide the scaffolding for knowledge graphs. For AI applications, ontologies improve search by understanding that related concepts should return relevant results (searching for 'analgesics' should find articles about paracetamol). They improve data integration by mapping equivalent terms across different systems. They improve AI accuracy by providing domain structure that guides model behaviour. Building and maintaining ontologies requires domain expertise and ongoing effort, but the investment pays dividends across multiple applications. A well-designed ontology can serve as the foundation for search, recommendation, compliance checking, data quality, and knowledge management systems.