Calibration Status

• CURIE: gmeow:CalibrationStatus
• IRI: https://blackcatinformatics.ca/gmeow/CalibrationStatus
• Category: class
• Defined by: gmeow:slices/metacognition
• Box roles: TBox role (What is this?)

The qualitative fit between an agent's avowed confidence and its actual accuracy — a closed value vocabulary: gmeow:wellCalibrated (confidence tracks accuracy), gmeow:overconfident (confidence exceeds accuracy — the Dunning–Kruger pole), and gmeow:underconfident (accuracy exceeds confidence — the impostor pole). Its members are individuals, never subclasses. The numeric magnitude of the gap is the solver-layer gmeow:calibrationError; this is the qualitative reading.

Structure

Subclass of: gufo:QualityValue

Practical Pattern

Use gmeow:CalibrationStatus as a specialized kind of gufo:QualityValue. Add statement metadata or a standpoint when the assertion needs provenance, confidence, or vantage.

Usage Advice

Use when

• Use as the value vocabulary for the qualitative calibration of a gmeow:MetacognitiveState — well-calibrated, over-, or under-confident — when the direction of the confidence/accuracy gap is what consumers route on.

Avoid when

• Avoid subclassing it to add a status (the vocabulary is closed and its members are individuals) and avoid using it to carry the numeric error magnitude (that is the solver-layer gmeow:calibrationError, not a value individual).

How to use

• Reference the three seeded gmeow:CalibrationStatus individuals from a gmeow:MetacognitiveState via gmeow:calibration; carry the numeric Brier-style gap, when computed, as a gmeow:calibrationError annotation on the asserting statement.

Examples

• ex:samDoubtsOwnGrasp gmeow:calibration gmeow:overconfident.