Simple TF-IDF

tf-idf.js

function getTermKey(term) {
  return term
    .normalize('NFD')
    .replace(/[\u0300-\u036f]/g, "")
    .replace(/[^\w]/g, "")
    .toLowerCase()
};

function getTermsIn(document) {
  return document.split(/[\s_():.!?,;]+/)
    .map(getTermKey);
}

function corpus() {
  const _documents = [];
  function add(document) {
    const termsInDocument = new Set(getTermsIn(document));
    _documents.push(termsInDocument);
  }

  function idf(term) {
    const termKey = getTermKey(term);
    const numberOfDocumentsContainingTerm = _documents.reduce((prev, document) => document.has(termKey) ? prev + 1 : prev, 0);
    return Math.log(_documents.length / (1 + numberOfDocumentsContainingTerm)) / Math.log(10);
  }

  function tf(term, document) {
    const termKey = getTermKey(term);
    const termsInDocument = getTermsIn(document)
    const termFrequency = termsInDocument.reduce((frequency, docTermKey) => docTermKey === termKey ? ++frequency : frequency, 0);
    return termFrequency / termsInDocument.length;
  }

  function tfIdf(term, document) {
    return tf(term, document) * idf(term);
  }

  return {
    add,
    tf,
    idf,
    tfIdf
  }
}