Interface: ExtractionStrategySchemaGuided
@kortexya/reasoninglayer / ExtractionStrategySchemaGuided
Interface: ExtractionStrategySchemaGuided
Section titled “Interface: ExtractionStrategySchemaGuided”Defined in: src/types/ingestion.ts:281
Schema-guided extraction strategy — NER + pattern-based, LLM only when needed.
Remarks
Section titled “Remarks”Wire format: {"type": "schema_guided", "ner_confidence_threshold": 0.7, "relation_confidence_threshold": 0.6, "escalate_unknown_types": true}
Properties
Section titled “Properties”escalateUnknownTypes
Section titled “escalateUnknownTypes”escalateUnknownTypes:
boolean
Defined in: src/types/ingestion.ts:288
Whether to escalate to LLM for unknown entity types.
nerConfidenceThreshold
Section titled “nerConfidenceThreshold”nerConfidenceThreshold:
number
Defined in: src/types/ingestion.ts:284
NER confidence threshold below which to escalate to LLM (0.0–1.0).
relationConfidenceThreshold
Section titled “relationConfidenceThreshold”relationConfidenceThreshold:
number
Defined in: src/types/ingestion.ts:286
Relation pattern confidence threshold (0.0–1.0).
type:
"schema_guided"
Defined in: src/types/ingestion.ts:282