/**
 * Standalone Deduplication Module
 *
 * Phase 3: Remove duplicate results from a batch of records.
 *
 * This is different from session fingerprinting (Phase 2):
 * - Session fingerprinting: tracks across multiple queries over time
 * - This module: dedupes a single batch (e.g., federated search output)
 *
 * Deduplication strategy (priority order):
 * 1. URL matching (primary) - exact match after normalisation
 * 2. Title similarity (fallback) - Jaccard coefficient threshold
 * 3. Year proximity - same title within ±N years
 */
import type { DedupeOptions, DedupeResult, DedupeMatchType } from './types.js';
/**
 * Normalise a URL for comparison
 * - Lowercase
 * - Remove trailing slashes
 * - Sort query parameters
 * - Remove fragments
 */
export declare function normaliseUrl(url: string): string;
/**
 * Normalise title for comparison
 * - Lowercase
 * - Remove punctuation
 * - Collapse whitespace
 */
export declare function normaliseTitle(title: string): string;
/**
 * Calculate Jaccard similarity between two titles
 * Returns a value between 0 (completely different) and 1 (identical)
 */
export declare function titleSimilarity(title1: string, title2: string): number;
/**
 * Check if two records are duplicates
 */
export declare function areDuplicates(a: Record<string, unknown>, b: Record<string, unknown>, options?: DedupeOptions): {
    isDuplicate: boolean;
    matchType?: DedupeMatchType;
};
/**
 * Deduplicate a batch of records
 *
 * @param records - Array of records to deduplicate
 * @param options - Deduplication options
 * @returns Deduplication result with unique records and statistics
 */
export declare function dedupeRecords(records: Record<string, unknown>[], options?: DedupeOptions): DedupeResult;
//# sourceMappingURL=dedupe.d.ts.map