Hadto note

Keet Notes · Chapter 5 · 2026-04-11

Competency questions need an authoring contract

Hadto’s latest ontology study pass points to a practical gap: a competency question is only operational if it is written in a form that can be turned into a repeatable check, query, or decision path.

Why this matters

This post shows how explicit models, workflow controls, and evidence trails make the business easier to inspect, teach, and run.

Why this note is here

Source check: Checks whether the source is useful before it shapes the work.

What supports it: Uses evidence, definitions, and cause-and-effect.

ontology engineeringcompetency questionshadtooperations

A lot of ontology work sounds more rigorous than it is because the language is formal while the intake is still loose.

Hadto’s latest reading pass through Chapter 5 Section 5.2.5 makes the gap plain. Writing competency questions is not enough. They have to arrive in a form the system can test, answer, and reuse.

Hadto is not building a library of tidy internal notes. It is building operating infrastructure for future owner-operators.

A question is not a contract

The usual workflow starts with a sensible idea: write down the questions the ontology should be able to answer. That gives direction, but it still leaves too much room for drift.

Two people can ask what looks like the same business question in different ways. An LLM can generate a polished question that sounds relevant but does not map cleanly to the ontology. A reviewer can accept a question because it feels important even though nobody has defined what counts as an answer. At that point, the competency question is still acting like a note.

What the platform can already do

The current Keet study tracker shows that Hadto already does some of the right things. It tracks competency questions as explicit artifacts, validates answer-path structure, filters malformed generated questions, and preserves lifecycle metadata around the questions it keeps.

Those pieces are a real base. The missing piece is an authoring contract. Contributors still are not told how to write a question so it becomes a reusable check instead of a one-off sentence.

What the contract should contain

The minimum version does not need to be heavy. It needs to be explicit.

A usable competency-question contract should include:

  • the question itself in controlled language
  • the business object or relation the question is about
  • the expected answer form, such as yes/no, list, count, threshold breach, or explanation
  • the intended executable target, such as a query, validation rule, report, or decision step
  • the owner or review role responsible for approving the question

For example: “Does every active location have an assigned general manager?” is not just a sentence. It is a yes/no question about active locations and manager assignment, it implies the entities and relation to inspect, and it points toward a query or validation check that can flag missing ownership before the issue turns into operational drift.

Why this matters for transfer

A loose question format breaks the handoff Hadto is trying to build. If competency questions stay free-form, every new contributor has to rediscover the unwritten rules: what belongs in the ontology, what vocabulary to use, what level of specificity is acceptable, and whether the result is meant to become a report, a check, a query, or a decision aid.

Scalable authoring cannot depend on that. It is institutional memory dressed up as methodology.

The same rule makes AI safer

The same contract also makes LLM-assisted ontology work safer.

If a model is allowed to generate free-form competency questions, it can produce plausible noise all day. If it has to fill a governed template with an expected answer form and executable destination, the job gets narrower and easier to review. The platform should not ask an AI to invent the meaning of the business. It should ask for candidate questions inside a defined contract.

The operating standard

Competency questions should enter the system with enough structure to survive handoff. If a contributor cannot name the answer form, target artifact, and review owner, the question is not ready. It is still a draft note.

That standard is simple. The sentence alone does not count.


Source evidence used in this note: smb-ontology-platform/docs/plans/2026-03-31-keet-ontology-engineering-progress-tracker.md (2026-04-11 entry for Chapter 5 Section 5.2.5) and existing Hadto blog posts reviewed to avoid duplicating prior notes on ontology methodology, pitfall scanning, and tooling discoverability. The upstream ONT-010 follow-through is referenced in the tracker but not yet available as a canonical public-facing issue document.

← Back to all notes