Theory
Adopting a standard is a dependency decision
Depending on a standard ontology is like depending on a software library: powerful, but you inherit its constraints. Evaluate it along these axes before you build on it:
- Licence & cost — Is it open (FIBO, OSDU, schema.org, OBO), affiliate-licensed (SNOMED CT needs a member/affiliate licence), or paywalled (ISO IDMP, ISO 15926 are purchased ISO documents)? Licensing decides whether you can even redistribute the terms in your product.
- Governance & cadence — Who maintains it, how often does it version, and how are change requests handled? A standard with no active steward is a liability.
- Expressivity / profile — Is it a lightweight vocabulary (schema.org, SKOS thesauri), a rich OWL 2 ontology (FIBO), or an OWL 2 EL terminology built for fast classification of huge hierarchies (SNOMED CT's OWL form)? The profile dictates what reasoning you can afford.
- Upper-ontology anchoring — Many standards anchor to a foundational ontology for consistency: BFO (OBO Foundry, ISO 15926-14/IDO) or a pragmatic upper like gist. Anchoring helps alignment later.
- Coverage & fit — Does it actually model your slice of the domain, or only 60% of it?
Two reuse strategies
- Import (
owl:imports) the standard and use its terms directly — maximal interoperability, but you inherit its full size and axioms. - Thin domain layer — reference the standard's IRIs and
rdfs:subClassOfthem from a small local ontology (the Atlas-Bank-on-FIBO pattern). You stay aligned without swallowing the whole import.
Use Case Example: A pharma team needs medicinal-product identity. IDMP fits, but the ISO texts are paywalled and huge, so they model a thin local layer whose classes rdfs:subClassOf the IDMP concepts and carry the regulator's required identifiers — aligned to the standard, without importing thousands of out-of-scope terms.