Catalog
concept#Data#Architecture#Governance#Integration

Ontology

An ontology defines formal concepts, their properties and relationships within a domain model. It provides a shared vocabulary for data integration, interoperability and semantic querying.

Ontologies are formal, machine-readable models that precisely describe concepts, relations and rules of a domain.
Established
High

Classification

  • High
  • Technical
  • Architectural
  • Advanced

Technical context

Knowledge graph and RDF repositoriesETL and data integration processesReporting and BI tools to consume semantic metadata

Principles & goals

Clear separation of terminology and implementation details.Iterative modelling in close collaboration with domain experts.Promote reusability of concepts across systems.
Discovery
Enterprise, Domain

Use cases & scenarios

Compromises

  • Overly complex models that reduce adoption and maintainability.
  • Incorrect or inconsistent term definitions cause inconsistencies.
  • Lack of governance leads to divergent extensions and silos.
  • Start with a lean core model and iterate.
  • Involve domain experts early and continuously.
  • Version and document changes clearly and traceably.

I/O & resources

  • Domain terms, glossaries and requirement documents
  • Sample data and source schemas
  • Access to subject-matter experts for clarifications
  • Machine-readable ontology models (e.g. OWL/RDF)
  • Mapping tables and transformation rules
  • Governance documents and versioning schemes

Description

Ontologies are formal, machine-readable models that precisely describe concepts, relations and rules of a domain. They harmonize data, enable semantic queries and improve interoperability across systems. Ontology engineering combines domain expertise with technical representations.

✔

  • Improved interoperability through unified terms.
  • Enables semantic queries and reasoning-based analyses.
  • Supports consistent metric and data interpretation.

✖

  • High initial effort for modelling and governance.
  • Maintenance effort when domain requirements change.
  • Not all data problems can be solved by ontologies alone.

  • Domain coverage

    Percentage of relevant terms covered by the ontology.

  • Number of reused concepts

    How many concepts are used across multiple systems or reports.

  • Validation errors per release

    Number of semantic inconsistencies or modelling errors after changes.

Schema.org (web ontology)

A widely used vocabulary for structuring web content and entities.

FOAF (person relationships)

A lightweight ontology model describing people and their relationships.

SNOMED CT (medical terminology)

An extensive standardized ontology for clinical terms and codings.

1

Inventory domain terms and schemas.

2

Create an initial ontology model with core entities.

3

Implement mappings and validation rules.

4

Establish governance and release procedures.

⚠️ Technical debt & bottlenecks

  • Unclear or non-versioned concept changes.
  • Ad-hoc mappings without a central standard.
  • Missing tests for ontology integrity after changes.
Lack of domain expertiseGovernance and versioningPerformance for complex reasoning operations
  • Using ontology as a substitute for poor data cleansing.
  • Overfitting the ontology to a specific tool format.
  • Uncontrolled free extensions by many teams.
  • Insufficient test cases for semantic validation.
  • Misunderstandings between technical and domain stakeholders.
  • Unconsidered performance requirements for reasoning.
Ontology modelling (OWL/RDF/Turtle)Domain knowledge and facilitationData integration and mapping techniques
Interoperability of heterogeneous data sourcesReusable domain terms and semanticsSupport for semantic queries and reasoning
  • • Existing data quality and semantics must be acceptable.
  • • Organisational agreement for shared vocabularies required.
  • • Platforms must support RDF/OWL or mappings.