/** Porter stem a lowercase token. Words ≤ 2 chars are returned unchanged. */
export declare function porterStem(word: string): string;
/**
 * Tokenise text into Lucene-style stemmed tokens: lowercase, split on
 * Unicode word boundaries, drop stopwords + length<2, Porter stem.
 */
export declare function luceneTokenize(text: string): string[];
export interface LuceneCorpusStats {
    df: Map<string, number>;
    idf: Map<string, number>;
    avgDocLen: number;
    N: number;
}
export declare function buildLuceneCorpusStats(tokenisedDocs: string[][]): LuceneCorpusStats;
/**
 * Lucene-style Okapi BM25: k1=1.2 (vs hybrid-retrieval's 1.5), b=0.75.
 */
export declare function luceneBM25(queryTokens: string[], docTokens: string[], stats: LuceneCorpusStats, k1?: number, b?: number): number;
//# sourceMappingURL=lucene-bm25.d.ts.map