Competency Questions — Your Spec, Test Suite and Scope Guard

Without CQs, your ontology has no scope

A competency question (CQ) is a natural-language question the ontology must be able to answer once populated. CQs are the cheapest, most powerful artefact in ontology engineering — they serve three roles at once:

1. Specification — they bound what's in scope.
2. Test suite — every CQ becomes a SPARQL ASK or SELECT.
3. Scope guard — anything the CQs don't justify, you don't build.

Anatomy of a good CQ

• Domain-language, not OWL-language. "Which orders did Alice place last quarter?" not "Find all :Order with :placedBy = :alice ⊓ :placedAt ∈ Q".
• Answerable by data, not by chat. "What's a good loyalty strategy?" is a CQ for a consultant, not an ontology.
• Numbered, traceable to one or more classes / properties in the model.

Aim for 20–50 CQs for a domain ontology. Less = under-spec. More = scope creep.

CQs are the acceptance tests of an ontology project.

In software, no one ships a feature without acceptance criteria; in ontology projects people routinely ship without competency questions and then wonder why the model doesn't do what stakeholders expected. Same disease, same cure: write the test first.