Embabel Agent Documentation/com.embabel.common.ai.decision/RatingResult

RatingResult

@ApiStatus.Experimental

The evidence a provider returned for a rating question: a selected level, a per-level distribution, a score, or any combination of these. Each part comes from the provider's response as reported. A selected level is present only when the provider selected one, and a distribution only when the provider reported one. Validation against a RatingQuestionSpec checks that a selected level and a distribution belong to the question's scale.

In JSON a rating result is an object whose status member names its class: answered, inconclusive or failure.

Inheritors

Answered

Inconclusive

Failure

Types

Answered

class Answered constructor(val provenance: ModelProvenance, val selectedLevelId: String? = null, distribution: List<LevelProbability> = emptyList(), val score: RatingScore? = null, val confidence: Double? = null) : RatingResult

A rating the provider answered with at least one piece of evidence: a selected level id, a distribution over levels, a score, or a combination of these. Confidence is the provider's own measure of how concentrated its reported distribution is. It is absent unless the provider actually reports it.

Failure

data class Failure(val reason: FailureReason) : RatingResult

An operational failure with no raw provider error or throwable retained.

Inconclusive

data class Inconclusive(val provenance: ModelProvenance) : RatingResult

Insufficient evidence to answer the rating.