Semi-structured sources still need a review gate

Semi-structured sources can suggest ontology terms, but they should never promote themselves into the business model. The review gate has to come first. A spreadsheet column, document attribute, or graph predicate is still source material until someone decides what it means, whether it is stable, and whether it belongs in shared ontology at all.

For Hadto, that rule matters because the company is trying to turn messy operating reality into reusable owner-operator infrastructure. When the platform treats flexible source fields as though they were already clean business terms, it formalizes local accidents instead of repeatable meaning.

Flexible sources still carry unstable meaning

Semi-structured data looks safer than it is. Document stores can seem closer to real-world nuance than rigid tables. Property graphs can look like ready-made domain structure. Spreadsheets feel approachable enough that domain experts may want to shape the model directly.

That flexibility is useful, but it lets several problems hide at once. The same idea can show up under different labels. Optional fields can imply unstable cardinality. Conflicting source assertions can coexist without a repair path. Column headings or node labels can mix business meaning with local shorthand.

A format can be easy to ingest and still be semantically unstable.

Candidate terms are still only candidates

Keet’s Chapter 7 sharpens an older Hadto lesson: a source term can suggest a concept, property, or workflow edge, but it does not prove one. Someone still has to decide what the term means, whether it is reusable, whether it belongs in the ontology or only in source data, and whether conflicting labels should be merged or rejected.

Without that review, the ontology becomes a cleaner-looking copy of source drift.

The gate has to sit before promotion

Downstream validation helps, but it is not the same as a review gate. If a semi-structured source is lifted into ontology form before contradiction review, duplicate handling, and cardinality review, the platform is already treating a source quirk as business structure. By the time validation runs, the system is checking a commitment that should still have been under review.

SHACL-style conformance is therefore not the same as deciding whether a candidate source assertion should have become ontology structure in the first place.

Spreadsheets need an authoring contract

Spreadsheets are the clearest example because they are easy to hand to an operator. That convenience only helps if the sheet has a declared contract.

Take a simple case: one branch manager adds a column called Customer Tier, another uses Account Class, and a third leaves the field blank for cash customers. Those columns may point to a useful business concept, but they may also reflect local reporting habits, temporary pricing logic, or incomplete data entry. The review gate is where Hadto decides whether those headings map to one shared concept, stay as source-only aliases, or should be rejected until the business rule is clearer.

The spreadsheet authoring contract should say which columns are candidate terms, which rows are allowed to propose structured assertions, which values remain examples or raw data, and which review step approves promotion into shared ontology artifacts. It should also leave a provenance trail showing what was accepted, merged, revised, or left out.

Absent that contract, the spreadsheet becomes another place where accidental vocabulary hardens into platform structure.

Why this matters to ownership

Hadto is not building ontology infrastructure as an academic exercise. The point is to make business systems transferable. An owner-operator should inherit a business model that reflects how the company actually works rather than a vendor export format, a team’s naming habits, or a one-off spreadsheet built under deadline pressure.

When semantic-lifting decisions stay implicit, operators inherit hidden modelling assumptions and weaker handoffs. The operating standard is simple: semi-structured sources can propose terms, but only reviewed terms get promoted into the ontology.

---

Source evidence used in this note: smb-ontology-platform/docs/plans/2026-03-31-keet-ontology-engineering-progress-tracker.md (2026-04-12 entries), smb-ontology-platform/docs/issues/ONT-026-add-semantic-lifting-governance-for-source-schema-classification-and-app-boundaries.md, and existing Hadto blog posts reviewed to avoid duplicating prior notes on semantic lifting, AI-assisted candidate generation, and relation-governance work.