/** * Extractor — Extracts structured knowledge entries from raw text. * * Paragraph-based splitting + classifier assignment. * Future: LLM-backed extraction with semantic chunking. * * Follows ADR-001 (pipeline filter) and ADR-005 (definition/execution separation): * Extractor defines extraction logic; actual LLM calls delegated to host in future. */ import type { KnowledgeEntry, KnowledgeSource } from '../types/index.js'; import { Classifier } from './classifier.js'; export interface ExtractorOptions { /** Minimum content length to consider a paragraph extractable */ minContentLength?: number; /** Custom classifier instance */ classifier?: Classifier; /** Minimum confidence threshold (retained for scoring; all entries are 'active') */ minConfidence?: number; } export declare class Extractor { private classifier; private minContentLength; private minConfidence; constructor(options?: ExtractorOptions); /** * Extract knowledge entries from raw text. * Splits by paragraphs, classifies each, returns structured entries. */ extract(text: string, source: KnowledgeSource): Promise; private splitIntoParagraphs; private generateTitle; private generateSummary; } //# sourceMappingURL=extractor.d.ts.map