Skip to content

Interface: FuzzyProveRequest

@kortexya/reasoninglayer


@kortexya/reasoninglayer / Inference / FuzzyProveRequest

Interface: FuzzyProveRequest

Defined in: src/types/inference.ts:390

Request for fuzzy proof search.

Remarks

Dual input mode: Provide EITHER goal OR goal_id, same as backward chaining.

Properties

goal?

optional goal: TermInputDto | null

Defined in: src/types/inference.ts:392

Inline goal. Mutually exclusive with goal_id.


goalId?

optional goalId: string | null

Defined in: src/types/inference.ts:394

Saved goal reference. Mutually exclusive with goal.


maxSolutions?

optional maxSolutions: number

Defined in: src/types/inference.ts:398

Maximum number of solutions.


minDegree?

optional minDegree: number

Defined in: src/types/inference.ts:396

Minimum degree threshold (0.0 to 1.0).


openWorld?

optional openWorld: boolean

Defined in: src/types/inference.ts:400

Whether to use open-world semantics.


saveGoal?

optional saveGoal: boolean

Defined in: src/types/inference.ts:402

Whether to persist the goal for later reuse.


scope?

optional scope: Record<string, string>

Defined in: src/types/inference.ts:416

Open-world reasoning scope: ground feature constraints every witnessing fact must satisfy, as feature → value (string values).

Remarks

Safety-relevant. For patient-scoped clinical reasoning send { patient_id: '<id>' } so only that patient’s observations witness a diagnosis — another patient’s fact of the same sort never does. A fact lacking the feature is unconstrained (open world). Empty or absent ⇒ tenant-wide (legacy) behavior, where any tenant fact may witness the goal.


tnorm?

optional tnorm: string

Defined in: src/types/inference.ts:404

T-norm strategy: “min”, “product”, or “lukasiewicz”.