frameworkExperience & interface

Ontology

An ontology formally defines the concepts in a domain, their properties, and the relationships and rules that connect them.

Download Markdown (.md)

01What it means

An ontology is an explicit model of a domain that describes classes of things, their attributes, their relationships, and sometimes constraints that allow new facts to be inferred. Unlike a simple taxonomy, which mainly classifies concepts, an ontology can represent richer statements such as a clinician treating a patient at a facility or a product being compatible with another product.[1][2][3]

A domain model richer than simple classification Classes gain properties, relationships, constraints, and instances that support inference
Domain questions
What must the model answer?
Classes
Kinds of things in the domain
Properties
AttributesIdentifiers
Relationships
DirectionCardinalityCompatibility
Constraints + rules
Allowed structuresInferred facts
Instances
Concrete entities linked by the model

Ontology depth should follow real competency questions; a lightweight taxonomy or schema is better when rich relationships and inference are unnecessary.

02Why it matters

Ontologies make meaning more consistent across systems and teams. They can support semantic search, knowledge graphs, data integration, recommendation, and machine reasoning by clarifying when differently named data refers to the same concept. The modeling work also exposes hidden assumptions that otherwise remain embedded in databases and interfaces.

03When to use it

Use an ontology when the product must combine complex data from several sources, represent many relationship types, or support queries that a category tree cannot answer. It is appropriate for domains such as healthcare, science, commerce, and asset management where shared semantics justify ongoing specialist governance.

04How it works

  1. Specify the domain, competency questions, intended users, and level of inference the model must support.
  2. Reuse established vocabularies where practical before creating new classes and properties.
  3. Model core entities, attributes, relationships, identifiers, and constraints with domain experts and data practitioners.
  4. Create representative instances and test whether the ontology can answer the competency questions without awkward exceptions.
  5. Publish documentation, ownership, mappings, and versioning rules so dependent systems can adapt safely.

05A practical example

Example

A cultural heritage platform models creators, works, physical objects, exhibitions, places, and time periods as distinct entities. A painting can be created by a person, held by an institution, shown in several exhibitions, and associated with a movement. Users can then explore relationships that a single folder hierarchy would hide.

06Where it breaks

Ontology work can become expensive and abstract, especially when teams model every theoretical distinction before proving a user need. Domain experts may disagree about definitions, and inference can produce surprising results when rules are incomplete. A lightweight schema or taxonomy is often better when the required relationships are simple.

Common mistakes

  • Treating an ontology as a more elaborate navigation tree.
  • Modeling the entire domain before validating a small set of valuable questions.
  • Creating local concepts when stable, interoperable vocabularies already exist.

07Sources

  1. OWL Web Ontology Language OverviewWorld Wide Web Consortium · W3C specification overviewOWL Web Ontology Language Overview
  2. OWL: Semantic Web StandardsWorld Wide Web Consortium · Standards hubOWL: Semantic Web Standards
  3. OWL 2 Web Ontology Language Document OverviewWorld Wide Web Consortium · W3C recommendationOWL 2 Web Ontology Language Document Overview
View saved topics

Start typing to search titles, definitions, categories, and frameworks.