Jump to text
On this page
Copy and source
Published source confirmed at last check

Source changed 2026-11-22 12:18:55 UTC · snapshot created 2026-11-22 12:19:56 UTC · last check 2026-11-22 12:45:10 UTC

F.17:12.6 - Model-comparison direction and level rows

Use these three rows to cite model refinement, its converse and the levels established by comparing models. B.5.MR gives the tests for the two model relations. The level row gives both directional readings of one order used in project documents and teaching.

In the shared ReferenceScheme a model’s completions satisfy its assertions. The inquiry and its interpretation determine the answer read from each completion, including probability claims when relevant. C refines A when at least one completion of C exists and every completion of C satisfies A under that interpretation. A is then an abstraction of C. Strictness requires an answer that A admits and C excludes. Within a chain established by this test, models with the same answers share a level for that inquiry. Models on different branches can give the same answer while imposing incompatible constraints.

For each row, UnificationThreadId is ModelComparisonDirections, GovernedValueKindRef is U.Kind and ReferenceScheme has the value given above. The stable row designators are retained. RowEditionId for this edition is 2026-11-18. Each sense cell gives its expression and local sense under the shared scheme.

UTSRowId: UTS.ModelRefinement.2026-11-16
GovernedValueRef: B.5.MR:4.2 refinement relation kind
SubjectPatternLocator: B.5.MR:4.1–4.2
UnifiedTechName: model refinement
NameCardRef: NC-MODEL-REFINEMENT
SenseCellRefs: SenseCell.ModelRefinement.2026-11-18
RowRationale: retain the specification-theory designation while making feasibility, interpretation and the converse recoverable
AdmissibleUse: cite C's feasible preservation of A for the named inquiry, including equality; use more concrete only with strictness
BlockedUse: the name alone establishes neither higher empirical accuracy nor a strict difference
CurrentnessCondition: reopen on changed predicate, interpretation, NameCard, cell or repeated confusion between refinement and accuracy

SenseCell.ModelRefinement.2026-11-18:
  LocalExpression: model refinement
  LocalSenseClaim: C refines A for inquiry Q under interpretation I exactly when at least one completion of C exists and every completion of C satisfies A under I; strictness additionally excludes an answer of A distinguishable for Q

UTSRowId: UTS.ModelAbstraction.2026-11-16
GovernedValueRef: B.5.MR:4.2 abstraction relation kind
SubjectPatternLocator: B.5.MR:4.2; B.5.MR:4.2.2
UnifiedTechName: model abstraction
NameCardRef: NC-MODEL-ABSTRACTION
SenseCellRefs: SenseCell.ModelAbstraction.2026-11-18
RowRationale: give the converse comparison a durable name while keeping model construction separate
AdmissibleUse: cite A as an abstraction of C under the same inquiry and interpretation by establishing that C refines A
BlockedUse: the name supplies no unique inverse construction and does not validate a wider empirical claim
CurrentnessCondition: reopen on changed predicate, interpretation, NameCard, cell or confusion between the relation and omitted visible detail

SenseCell.ModelAbstraction.2026-11-18:
  LocalExpression: model abstraction
  LocalSenseClaim: A is an abstraction of C for inquiry Q under interpretation I exactly when C refines A for Q under I. Calling A more abstract than C requires strictness for Q.

UTSRowId: UTS.ModelAbstractionLevel.2026-11-16
GovernedValueRef: B.5.MR:4.3 inquiry-relative level of model abstraction
SubjectPatternLocator: B.5.MR:4.2–4.3
UnifiedTechName: level of model abstraction
NameCardRef: NC-MODEL-ABSTRACTION-LEVEL
SenseCellRefs: SenseCell.ModelAbstractionLevel.2026-11-18, SenseCell.ModelRefinementLevel.2026-11-18
RowRationale: name the ordered levels in both directions so the reader can identify which models are compared and what follows from the comparison
AdmissibleUse: read L0 towards L4 as levels of model abstraction and L4 towards L0 as levels of model refinement in an established application chain. Use intermediate levels where the comparison warrants them.
BlockedUse: a level name alone establishes neither preservation nor a strict difference for the inquiry
CurrentnessCondition: reopen on changed comparison, NameCard, cell or repeated use as a global rank

SenseCell.ModelAbstractionLevel.2026-11-18:
  LocalExpression: level of model abstraction
  LocalSenseClaim: level in an established refinement chain read towards a model that allows an answer the other model excludes for Q under I. Successive models with the same answers share a level for Q.

SenseCell.ModelRefinementLevel.2026-11-18:
  LocalExpression: level of model refinement
  LocalSenseClaim: the same order read towards a model that excludes a previously admissible answer for Q under I. Successive models with the same answers share a level for Q.

For a source that says “level”, “scale” or “layer”, use these model rows when it compares models for one inquiry. Library.Reference:6.2.2 gives the companion F.17 rows for the composition of each kind of holon, functional layers, measurement scales and the scope kinds of USM. A Method stack can be read for both its functional organization and its constituent Methods where both relations are established. To recover the range of a speaker’s “all” before choosing a row, use C.39.QR:4.1.1.

The rows cite no F.9 Bridge: they make no correspondence claim between different local meanings. A concrete comparison still needs its models, inquiry, interpretation and B.5.MR test; citing a row supplies the selected name and its meaning.