Green reasoners are not enough: RBox governance is the next ontology gate

Chapter 5 draws an operational distinction Hadto should keep front and center: consistency is not the same as semantic quality.

For an ontology-driven venture platform, that gap is not academic. It is the difference between a system that routes business logic reliably and one that looks clean in dashboards while relation semantics drift underneath it.

What Hadto checks today

The current stack already runs class-level logical checks. Evidence in tools/validate_logical.py and server/scripts/validation_service.py points to checks such as:

• subclass closure,
• unsatisfiable-class detection,
• consistency checking.

That is a good baseline. It tells us whether the class hierarchy is contradictory and whether the model is formally coherent at that level.

What those checks miss

Keet’s point is that relation-level modelling can be wrong without creating a contradiction.

A subproperty can be mis-scoped. A domain/range pairing can be semantically off. A property chain can combine relations in ways that imply inferences the business does not want. None of that is guaranteed to break class consistency checks.

That is the current Hadto gap. The study evidence says tools/logical_reasoning.py focuses on subclass closure and unsatisfiability paths, but does not inspect rdfs:subPropertyOf compatibility, object-property narrowing, or property-characteristic safety. The public logical surface mainly reports class-level services, with no dedicated RBox audit payload.

At the same time, Hadto already carries active relation commitments, including infer:promotionPeer a owl:SymmetricProperty in shared inference rules. Relation semantics are already shaping reasoning behavior. They are not hypothetical.

Current governance text also flags owl:propertyChainAxiom as unsupported drift in parts of the reasoning profile. That is a useful warning, but it does not answer the question that matters in operations: is a given property hierarchy semantically safe to use in this domain, even if it stays internally consistent?

Why this is a business problem

Hadto wants ontology inferences to support repeated operational decisions. If relation semantics are wrong but still consistent, the platform can make confidently wrong suggestions:

• role edges may imply broader permissions than intended,
• workflow links may cross contexts inappropriately,
• ownership or association inferences may get noisy while still passing checks.

That is a bad failure mode for growth infrastructure because the system still looks healthy.

What green should mean

Hadto should separate two risk classes:

• logical inconsistency risk: contradictions and unsatisfiable classes,
• semantic drift risk: relation definitions that are coherent but wrong for domain operations.

The first is already partly observable. The second is not yet first-class in production validation output.

That is why the study run opened ONT-007: to add a practical RBox and property-chain audit path that can warn when relations are mathematically coherent but semantically suspect. That is not optional polish. It is quality infrastructure for a platform that depends on high-signal meaning in ownership, obligation, and workflow relations.

Practical next step

The needed move is straightforward:

• add explicit checks for RBox compatibility,
• keep diagnostics structured instead of pass/fail only,
• point warnings to the exact relation path that is fragile,
• tie those findings back to ORSD and validation governance before they leak into product behavior.

A green logical validation should mean more than “no contradictions.” It should move toward “no semantically unsafe relation structures” as well.