What is an Ontology?

An introduction to ontologies in computer science and the Semantic Web.

Definition

In philosophy, an ontology is the study of what exists — the nature of being and the categories of reality. In computer science and the Semantic Web, the term has been borrowed to mean something more specific: a formal, explicit specification of a shared conceptualization.

Put simply, an ontology is a structured way to describe the concepts in a domain (such as medicine, law, or cultural heritage) and the relationships between them. Unlike a simple glossary or database schema, an ontology uses logic-based formalisms that allow machines to reason about the data automatically.

Why Ontologies Matter

Interoperability: Ontologies provide a common vocabulary so that different systems, organizations, and datasets can share and integrate data without ambiguity.
Reasoning: Because ontologies are logic-based, reasoners can infer new facts that were not explicitly stated — for example, deducing that a “Penguin” is a “Bird” even if only the subclass chain is defined.
Knowledge reuse: Well-designed ontologies can be shared and extended across projects, preventing each team from reinventing the same domain model.
Entailment & consistency checking: Ontologies express constraints (e.g., a person has at most one birth date) that a reasoner uses to derive new facts and detect contradictions. Because OWL adopts the open-world assumption, it isn't a closed-world data validator — for that, use SHACL or a similar shape/constraint language alongside the ontology.

Real-World Use Cases

Biomedical Sciences

The Gene Ontology (GO) and SNOMED CT organize biological processes and clinical terminology used by researchers and hospitals worldwide.

Cultural Heritage

CIDOC-CRM models museum artifacts, historical events, and provenance so that collections across institutions can be linked together.

E-Commerce

Schema.org provides structured product data that search engines use to display rich snippets and power product comparisons.

Libraries & Publishing

Dublin Core and BIBFRAME describe publications and library holdings, enabling cross-catalog search and linked open data.

The Semantic Web Stack

Ontologies sit at the heart of the W3C Semantic Web architecture, often visualized as a layered “cake”:

  ┌───────────────────────┐
  │    Trust / Proof      │
  ├───────────────────────┤
  │    Rules (SWRL)       │
  ├───────────────────────┤
  │  Ontologies (OWL)     │  ← you are here
  ├───────────────────────┤
  │  Vocabularies (RDFS)  │
  ├───────────────────────┤
  │  Data Model (RDF)     │
  ├───────────────────────┤
  │  Identifiers (URIs)   │
  ├───────────────────────┤
  │  Syntax (XML / JSON)  │
  └───────────────────────┘

Each layer builds on those below it. RDF provides the basic data model of triples (subject, predicate, object). RDFS adds vocabulary constructs like classes and properties. OWL adds formal logic for richer ontology modeling.

Key Terms Glossary

Term	Definition
Class	A category or type of thing (e.g., Person, Animal, Disease).
Property	A relationship between entities (e.g., hasParent) or between an entity and a value (e.g., hasAge).
Individual	A specific instance of a class (e.g., “Albert Einstein” is an individual of class Person).
IRI	Internationalized Resource Identifier — the globally unique name for any resource in an ontology.
Triple	The atomic unit of RDF data: subject – predicate – object (e.g., “Einstein – bornIn – Germany”).
Reasoner	Software that infers new knowledge from the axioms and assertions in an ontology.