accepted 2026-05-31

Gated gUFO `rdf:type` Property-Typing Pass

Context and Problem Statement

Council session-029 (Full panel) judged the ODR-0008 §Q2a(a) UFO-axis spawn trigger NOT met (Q1–Q4 each 0–6–0) and declined to mint ODR-0008a/b/c: the curated walk typed leaves by bearer Kind (rdfs:domain), not UFO meta-category, and the flagship leaves fail OntoClean as a partition backbone (priceQualifier straddles Mode/Quality; ownershipType rides two identity principles — Guarino). But the panel unanimously affirmed (Q5, 6–0–0) that the Quality/Mode distinction is ontologically real and rigid as a typing (Guizzardi, Guarino, Knublauch, Davis) and ruled it should be preserved via a per-property gUFO rdf:type annotation — “the distinction earns rdf:type; it does not earn a namespace.” Verified live, no gufo: prefix and not one rdf:type gufo:Quality/gufo:Mode triple exists in the corpus today — the typing pass was never run. This work item is logged at ADR-0005 §G25 (queued from session-029 Q5); running it is conjunct (i) of the ODR-0023 R2 re-open trigger. This ADR is the engineering record for the emission; the modelling decision and the per-leaf identity criterion remain owned by ODR-0008 §Q5a (session-029 §“No new ODR”).

Decision Drivers

session-029 Q5 (6–0–0) — the ratified disposition: a gated gUFO rdf:type typing pass over the uncontested Property descriptive leaves, preserving the insight without a namespace.
ODR-0010 §Q7a — Knublauch’s pin: the typing is a classification triple, not a constraint, so it lives in the annotation graph, never the shapes graph (CI-test-3: ASK { GRAPH opda:annotations { ?s a sh:NodeShape } } → FALSE). Not SHACL (no validator exists; that re-incurs the rejected structure cost).
Uncontested-only — type the clean Quale-in-Region leaves; omit the straddlers (priceQualifier, marketingTenure) and the re-sorter (ownershipType) until a rigid rule adjudicates each to exactly one cell (Guarino’s identity/disjointness gate). tenureKind is a Substance-Kind label, not a Quality/Mode — also omitted.
Classification, not axiom — the rdf:type is an advisory UFO meta-category marker on the attribute, quarantined to the annotation graph so no reasoner reads the datatype property as a quality individual; the bearer’s identity criterion stays ODR-0008 §Q5a’s. No new bearer, no fresh IC (session-029 line 134).
Reversible / subtractive posture (S025/S029) — assert a rdf:type over a w3id namespace; do not mint permanent modules around an analytic lens that re-sorts under the next reasoner pass.

Considered Options

Per-property rdf:type gufo:Quality in the annotation graph — the dominant position (Guizzardi/Guarino/Knublauch/Davis); exactly as the verdict frames it.
skos:scopeNote / rdfs:comment prose only — Allemang’s conservative minority on mechanism (not route): keep prose until a typing pass + a biting query arrive, lest the assertion manufacture an entailment no query needs. Recorded, non-blocking.
SHACL over the annotation, or a UFO sub-module namespace (ODR-0008a/b/c) — rejected by session-029 (no validator; partition backbone fails OntoClean on the straddlers).

Decision Outcome

Chosen option: per-property rdf:type gufo:Quality in the descriptive annotation graph, gated to the uncontested leaves — the unanimous Q5 route. Emitted via the ADR-0030 generator (no hand-edited TTL), extending the ADR-0012 annotation emitter (emitters/annotations.py::build_descriptive_annotations):

Prefix — declare gufo: <http://purl.org/nemo/gufo#>, bound only in opda-descriptive-annotations.ttl (the other five annotation files are byte-unchanged).
Typed leaves (5) — gufo:Quality rdf:type on every §Q5a Quale-in-Region leaf on opda:Property: opda:currentEnergyRating (EPC), opda:councilTaxBand (council-tax), opda:builtForm (built-form), opda:centralHeatingFuelType, opda:heatingType. Each also carries dct:source <…/ODR-0008#section-Q5a>.
Omitted (adjudicated-pending / out-of-category) — opda:ownershipType (re-sorter: quality-by-type vs legal-estate-by-bearer); opda:priceQualifier, opda:marketingTenure (Mode/Quality straddlers — and descriptive-layer, not core Property); opda:tenureKind (Substance-Kind label). A module-level rdfs:comment records the inclusion rule + the omissions + the advisory (non-axiom) reading.
Home discipline — the triples land in opda-descriptive-annotations.ttl (the advisory annotation graph), never in opda-classes.ttl, opda-inference.ttl, or any *-shapes.ttl. No gufo:Quality is declared as an owl:Class here (it is referenced, reference-not-import, like DPV).
NOT emitted: any gufo:Mode (no uncontested Mode leaf exists — the Modes are the omitted straddlers); any SHACL; any ODR-0008a/b/c namespace; any owl:imports <…/gufo>.

Consequences

Good, because it preserves the UFO Quality insight the panel affirmed, as a queryable rdf:type, without minting the rejected modules.
Good, because it builds conjunct (i) of the ODR-0023 R2 re-open trigger — a future honest spawn judgement now has a typed leaf-set to test a biting query against (conjunct (ii)).
Good, because it incurs no namespace, no IC, no key debt and is reversible (a classification triple in the advisory graph retracts cleanly; re-homes onto a sub-module if one is ever earned).
Neutral, because the straddlers/re-sorter stay untyped by design — adjudicating priceQualifier (Guizzardi: a Quality of the listing Relator) / ownershipType to one cell is a separate, named follow-on, not blocked here.
Neutral, because Allemang’s prose-until-a-biting-query minority is recorded — if the typing is later judged to manufacture an unused entailment, it is annotation-graph-isolated and removable without touching classes/shapes.
Bad (accepted), because asserting rdf:type gufo:Quality on a datatype property is a deliberate lightweight abbreviation of “this attribute is of the Quality meta-category”, not a literal claim that the property is a quality particular — mitigated by the annotation-graph quarantine + the module rdfs:comment that states the reading.

Confirmation

tests/test_annotations.py::test_descriptive_gufo_quality_typing asserts the 5 leaves carry (leaf, rdf:type, gufo:Quality) + dct:source ODR-0008 §Q5a, and that the 4 omitted leaves do not.
ci-three-graph (check_no_shacl_in_annotations) stays green — the typing uses rdf:type/dct:source, no sh:*; test_no_owl_class_triples_in_annotations stays green (no ?s a owl:Class).
ci-byte-identity re-pins only opda-descriptive-annotations.ttl (the other five annotation files + all class/shape graphs are byte-unchanged); test_annotation_files_byte_identical_across_runs confirms determinism.
ci-category-g-coverage unchanged at 239/239 (this is a typing annotation, not a new leaf); full opda-gen pytest suite green.

More Information

Modelling decision + IC owner: ODR-0008 §Q5a (per-leaf Quale-in-Region bindings) + §Q2a(a) (the spawn-gate refinement); Council session-029 Q5 (the 6–0–0 disposition + DA scorecard + the held-as-live pro-spawn dissent).
Scheduling + scope: Council session-029 Q5 (the 6–0–0 disposition) + ADR-0005 §G25 (the queued work item — pure-additive and sound, executed here); ODR-0023 R2 (conjunct (i) of the re-open trigger).
Graph-boundary contract: ODR-0010 §Q7a (annotation-vs-shapes graph); ODR-0004 §3a (three-graph separation).
Deferred-work register: ADR-0005 §G25 (this work item, now executed).
Generator: ADR-0012 (the annotation emitter this extends); ADR-0030 (the descriptive emission subsystem).
gUFO vocabulary: Almeida, Guizzardi et al., gUFO: A Lightweight Implementation of UFO (NEMO, 2019), http://purl.org/nemo/gufo#.
AgentDB registration of this ADR is pending an adr-index run (the ruflo MCP is disconnected this session); the file + frontmatter edges are authoritative.

Gated gUFO rdf:type Property-Typing Pass