calibration error
- CURIE:
gmeow:calibrationError - IRI: https://blackcatinformatics.ca/gmeow/calibrationError
- Category: property
- Defined by:
gmeow:slices/metacognition - Box roles: RBox role (What is this?)
A SOLVER-LAYER numeric measure of miscalibration (Principle 12) — a Brier-style score in the closed interval [0,1] attached to the statement asserting a metacognitive calibration, quantifying the gap between avowed confidence and accuracy (0 = perfectly calibrated). An owl:AnnotationProperty by design, exactly like gmeow:confidence: it is INVISIBLE to the reasoner and can never become a materialised axiom, so the calibration metric stays computed-not-asserted. The qualitative direction is gmeow:calibration; this is the magnitude.
Structure
Property shape: annotation property
Practical Pattern
Use gmeow:calibrationError when the definition matches the source fact. Prefer a narrower GMEOW term when one exists, and keep projection concerns in the mapping layer.
Example Snippets
These snippets are generated from canonical slice examples and trimmed to the Turtle blocks where this term appears.
Dunning Kruger
- Source:
slices/core/metacognition/examples/dunning-kruger.ttl - Examples catalog: open in catalog#example-slices-core-metacognition-examples-dunning-kruger
# SPDX-FileCopyrightText: 2026 Blackcat Informatics® Inc. <paudley@blackcatinformatics.ca>
# SPDX-License-Identifier: CC-BY-4.0
#
# Worked example — Dunning–Kruger as a MODELLED metacognitive miscalibration, the
# second-order companion to the cognition slice's dunning-kruger.ttl. Where that
# example records two coexisting first-order gmeow:KnowledgeProficiency claims at
# divergent depths (Principle 9), this one turns the reflexive lens on Sam's OWN
# self-assessment: a gmeow:MetacognitiveState whose gmeow:metaTarget is Sam's
# self-attributed mastery, carrying gmeow:calibration gmeow:overconfident. The
# numeric Brier-style gap rides gmeow:calibrationError — a SOLVER-LAYER annotation
# (Principle 12), invisible to the reasoner. The calibration is itself a
# vantage-indexed claim (Principle 9): the assessor judges Sam overconfident; GMEOW
# records the judgement, it does not adjudicate a global verdict.
@prefix gmeow: <https://blackcatinformatics.ca/gmeow/> .
@prefix ex: <https://blackcatinformatics.ca/gmeow/examples/metacognition/> .
@prefix wd: <http://www.wikidata.org/entity/> .
# --- The calibration is a vantage-indexed claim (Principle 9): the assessor,
# observing Sam's self-regard against a knows-about assessment, judges it
# overconfident and records the Brier-style magnitude as a solver-layer
# annotation on the claim (computed-not-asserted, Principle 12).
ex:samCalibrationClaim a gmeow:StandpointClaim ;
gmeow:vantage ex:assessor ;
gmeow:observedFeature ex:samSelfRegard ;
gmeow:observationMethod gmeow:methodExpertJudgement ;
gmeow:claimModality gmeow:probable ;
gmeow:calibrationError 0.62 .
Usage Advice
Use when
- Use to attach a solver-computed Brier-style miscalibration magnitude in [0,1] to the statement that asserts a metacognitive calibration, when the numeric error is worth recording alongside the qualitative
gmeow:calibration.
Avoid when
- Avoid modelling it as an owl:DatatypeProperty with a domain — that would make it a reasoned ABox edge and risk the reasoner touching it, violating the solver boundary (Principle 12) — and avoid deriving any entailment from it; it is recorded structure, not logic.
How to use
- Attach
gmeow:calibrationErroras an annotation on the calibration statement (the reified claim or the triple), carrying the externally-computed Brier-style value in [0,1]; leave its computation entirely to the solver layer.
Examples
- ex:samCalibrationClaim
gmeow:calibrationError0.62.