cmpstr/dist/esm/metric/Jaccard.mjs

CmpStr is a lightweight, fast and well performing package for calculating string similarity

// CmpStr v3.2.2 build-bb61120-260311 by Paul Köhler @komed3 / MIT License
import { Pool } from '../utils/Pool.mjs';
import { MetricRegistry, Metric } from './Metric.mjs';

class JaccardIndex extends Metric {
  constructor(a, b, opt = {}) {
    super('jaccard', a, b, opt, true);
  }
  compute(a, b, m, n) {
    const [setA, setB] = Pool.acquireMany('set', [m, n]);
    try {
      for (const A of a) setA.add(A);
      for (const B of b) setB.add(B);
      let intersection = 0;
      for (const c of setA) if (setB.has(c)) intersection++;
      const union = setA.size + setB.size - intersection;
      return {
        res: union === 0 ? 1 : Metric.clamp(intersection / union),
        raw: { intersection, union }
      };
    } finally {
      Pool.release('set', setA, m);
      Pool.release('set', setB, n);
    }
  }
}
MetricRegistry.add('jaccard', JaccardIndex);

export { JaccardIndex };