@thi.ng/text-analysis/similarity.js

Text tokenization, transformation & analysis transducers, utilities, stop words, porter stemming, vector encodings, similarities

import { distCosine } from "@thi.ng/vectors/dist-cosine";
import { jaccardSimilarity as $jaccard } from "@thi.ng/vectors/dist-jaccard";
import { distSq as $distSq } from "@thi.ng/vectors/distsq";
import { dot as $dot } from "@thi.ng/vectors/dot";
const cosineSimilarityDense = distCosine;
const cosineSimilaritySparse = (a, b) => {
  const dot = dotProductSparse(a, b);
  return dot > 0 ? dot / (Math.sqrt(a.length) * Math.sqrt(b.length)) : 0;
};
const jaccardSimilarityDense = $jaccard;
const jaccardSimilaritySparse = (a, b) => {
  const na = a.length;
  const nb = b.length;
  let numIsec = 0;
  let apos, bpos;
  for (let i = 0, j = 0; i < na && j < nb; ) {
    apos = a[i];
    bpos = b[j];
    if (apos === bpos) {
      numIsec++;
      i++;
      j++;
    } else if (apos < bpos) {
      i++;
    } else {
      j++;
    }
  }
  const numUnion = na + nb - numIsec;
  return numUnion > 0 ? numIsec / numUnion : 0;
};
const dotProductDense = $dot;
const dotProductSparse = (a, b) => {
  let res = 0;
  let apos, bpos;
  for (let i = 0, j = 0, na = a.length, nb = b.length; i < na && j < nb; ) {
    apos = a[i];
    bpos = b[j];
    if (apos === bpos) {
      res++;
      i++;
      j++;
    } else if (apos < bpos) i++;
    else j++;
  }
  return res;
};
const distSqDense = $distSq;
const distSqSparse = (a, b) => {
  let res = 0;
  let apos, bpos;
  for (let i = 0, j = 0, na = a.length, nb = b.length; i < na && j < nb; ) {
    apos = a[i];
    bpos = b[j];
    if (apos === bpos) {
      i++;
      j++;
    } else {
      res++;
      if (apos < bpos) i++;
      else j++;
    }
  }
  return res;
};
export {
  cosineSimilarityDense,
  cosineSimilaritySparse,
  distSqDense,
  distSqSparse,
  dotProductDense,
  dotProductSparse,
  jaccardSimilarityDense,
  jaccardSimilaritySparse
};