What is Ontology?

Without an ontology, data integration is a brute-force mapping exercise that breaks every time a source system changes. An ontology provides a stable, business-aligned vocabulary that decouples meaning from implementation.

When multiple teams, departments, or organisations agree on a shared ontology, they can exchange and combine data without costly point-to-point integrations. This is especially valuable in regulated industries where consistent terminology is a compliance requirement.

Ontologies also enable AI systems to reason: if the ontology states that “Paracetamol is a type of Analgesic” and “Analgesics treat Pain”, a machine can infer that Paracetamol treats Pain - without being explicitly told.

An ontology is typically expressed in OWL (Web Ontology Language) or RDFS (RDF Schema). It consists of:

Classes: The types of things in the domain (e.g., Person, Organisation, Product).

Properties: The attributes and relationships (e.g., hasName, worksFor, partOf).

Constraints: Rules such as cardinality (a Person has exactly one date of birth) and domain/range restrictions (worksFor links a Person to an Organisation).

Ontology engineering is the disciplined process of designing, validating, and evolving these models. It involves domain experts, data architects, and ontology engineers working together to capture knowledge accurately and pragmatically.

FIBO (Financial Industry Business Ontology): A community-standard ontology for financial services covering entities, instruments, and regulatory concepts. Banks use FIBO to harmonise data across trading, risk, and compliance systems.

Schema.org: A lightweight ontology used by millions of websites to mark up structured data for search engines. When you see rich snippets in Google, Schema.org ontology is at work.

Enterprise product catalogues: A manufacturer uses a custom ontology to classify 500,000 SKUs across divisions, enabling cross-selling, consistent search, and automated regulatory reporting.