data-sampler

/**
 * @file priority.ts
 * @description Priority-based sampling strategy.
 * Scores chunks by keyword frequency (auto-extracted from the text), position
 * (earlier content slightly boosted), and chunk density. Returns chunks sorted
 * by score descending.
 */
import type { Chunk, SamplingConfig } from "../types";
/**
 * Score chunks by keyword frequency and position.
 *
 * Algorithm:
 * 1. Extract top keywords from the full text (by frequency).
 * 2. Split text into fixed-size chunks using the length-based strategy.
 * 3. Score each chunk by keyword density (primary) and position (secondary boost).
 * 4. Return chunks sorted by score descending (highest priority first).
 *
 * @param text - The input text to sample.
 * @param config - Sampling configuration (chunkSize, overlap, maxChunks).
 * @returns Chunks sorted by priority score descending.
 */
export declare function sampleByPriority(text: string, config: SamplingConfig): readonly Chunk[];
//# sourceMappingURL=priority.d.ts.map