is_organisation.js

Jump To … +

abbreviations.js adjectives.js convertables.js dates.js demonyms.js firstnames.js honourifics.js irregular_nouns.js irregular_verbs.js misc.js multiples.js numbers.js organisations.js phrasal_verbs.js places.js uncountables.js verbs.js fns.js index.js lexicon.js negate.js passive_voice.js contractions.js fancy_lumping.js grammar_rules.js parts_of_speech.js phrasal_verbs.js tagger.js word_rules.js question.js sentence.js statement.js tense.js adjective.js to_adverb.js to_comparative.js to_noun.js to_superlative.js adverb.js to_adjective.js is_acronym.js article.js date.js date_rules.js is_date.js parse_date.js is_plural.js is_uncountable.js noun.js is_organisation.js organisation.js gender.js is_person.js parse_name.js person.js is_place.js place.js pluralize.js pronoun.js singularize.js is_value.js numbers.js to_number.js units.js value.js term.js conjugate.js from_infinitive.js predict_form.js suffix_rules.js to_actor.js to_infinitive.js negate.js verb.js sentence_parser.js text.js

is_organisation.js

'use strict';
const abbreviations = require('../../../data/abbreviations');
const org_data = require('../../../data/organisations');

words like ‘co’ and ltd

let org_suffix = abbreviations.orgs.reduce(function(h, s) {
  h[s] = true;
  return h;
}, {});
org_data.suffixes.forEach(function(s) { //a few more
  org_suffix[s] = true;
});

named orgs like google and nestle

let org_names = org_data.organisations.reduce(function(h, s) {
  h[s] = true;
  return h;
}, {});

const is_organisation = function(str, text) {
  text = text || '';

some known organisations, like microsoft

  if (org_names[str]) {
    return true;
  }

no period acronyms

  if (text.length <= 5 && text.match(/^[A-Z][A-Z]+$/) !== null) {
    return true;
  }

period acronyms

  if (text.length >= 4 && text.match(/^([A-Z]\.)*$/) !== null) {
    return true;
  }

eg ‘Smith & Co’

  if (str.match(/ & /)) {
    return true;
  }

Girlscouts of Canada

  if (str.match(/..s of /)) {
    return true;
  }

eg pets.com

  if (str.match(/[a-z]{3}\.(com|net|org|biz)/)) { //not a perfect url regex, but a "org.com"
    return true;
  }
  let words = str.split(' ');
  let last = words[words.length - 1];
  if (org_suffix[last]) {
    return true;
  }

  return false;
};

module.exports = is_organisation;

¶

console.log(is_organisation(‘Captain of Jamaica’));