Interface: LtnRefuteRequest

@kortexya/reasoninglayer / LTN / LtnRefuteRequest

Interface: LtnRefuteRequest

Defined in: src/types/ltn.ts:243

Request to verify whether candidateRules is a logical consequence of kbRules at satisfaction threshold q (reasoning by refutation).

Example

const request: LtnRefuteRequest = {
  kbRules: [{ individual: 'rex', sort: 'dog', membership: 0.9 }],
  candidateRules: [{ individual: 'rex', sort: 'animal', membership: 0.95 }],
  q: 0.7,
};

Properties

aggregator?

optional aggregator: LtnAggregator | null

Defined in: src/types/ltn.ts:249

KB-level aggregator. Absent ⇒ pMeanError(p = 2).

candidateRules

candidateRules: LtnRule[]

Defined in: src/types/ltn.ts:247

The candidate formula φ.

gridSteps?

optional gridSteps: number | null

Defined in: src/types/ltn.ts:254

Certainty-grid resolution per rule for the search.

Default Value

kbRules

kbRules: LtnRule[]

Defined in: src/types/ltn.ts:245

The knowledge base K.

q?

optional q: number | null

Defined in: src/types/ltn.ts:259

Satisfaction threshold q ∈ (0.5, 1).

Default Value

0.7

refineSteps?

optional refineSteps: number | null

Defined in: src/types/ltn.ts:264

Local-refinement steps per rule.