cmpstr/dist/cjs/metric/Cosine.cjs

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
'use strict';

var Pool = require('../utils/Pool.cjs');
var Metric = require('./Metric.cjs');

class CosineSimilarity extends Metric.Metric {
  constructor(a, b, opt = {}) {
    super('cosine', a, b, opt, true);
  }
  _termFreq(str, delimiter) {
    const terms = str.split(delimiter);
    const freq = Pool.Pool.acquire('map', terms.length);
    for (const term of terms) freq.set(term, (freq.get(term) || 0) + 1);
    return freq;
  }
  compute(a, b) {
    const { delimiter = ' ' } = this.options;
    const termsA = this._termFreq(a, delimiter);
    const termsB = this._termFreq(b, delimiter);
    try {
      let dotP = 0,
        magA = 0,
        magB = 0;
      for (const [term, freqA] of termsA) {
        const freqB = termsB.get(term) || 0;
        dotP += freqA * freqB;
        magA += freqA * freqA;
      }
      for (const freqB of termsB.values()) magB += freqB * freqB;
      magA = Math.sqrt(magA);
      magB = Math.sqrt(magB);
      return {
        res: magA && magB ? Metric.Metric.clamp(dotP / (magA * magB)) : 0,
        raw: { dotProduct: dotP, magnitudeA: magA, magnitudeB: magB }
      };
    } finally {
      Pool.Pool.release('map', termsA, termsA.size);
      Pool.Pool.release('map', termsB, termsB.size);
    }
  }
}
Metric.MetricRegistry.add('cosine', CosineSimilarity);

exports.CosineSimilarity = CosineSimilarity;