• 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

  • lexicon.js

  • ¶

    the lexicon is a big hash of words to pos tags it’s built by conjugating and inflecting a small seed of terms

    'use strict';
const fns = require('./fns.js');
const verb_conjugate = require('./term/verb/conjugate/conjugate.js');
const to_comparative = require('./term/adjective/to_comparative.js');
const to_superlative = require('./term/adjective/to_superlative.js');
const grand_mapping = require('./sentence/pos/parts_of_speech.js').tag_mapping;

const lexicon = {};

const addObj = function(obj) {
  const keys = Object.keys(obj);
  const l = keys.length;
  for (let i = 0; i < l; i++) {
    lexicon[keys[i]] = obj[keys[i]];
  }
};

const addArr = function(arr, tag) {
  const l = arr.length;
  for (let i = 0; i < l; i++) {
    lexicon[arr[i]] = tag;
  }
};
  • ¶

    conjugate all verbs.

    const verbMap = {
  infinitive: 'Infinitive',
  present: 'PresentTense',
  past: 'PastTense',
  gerund: 'Gerund',
  actor: 'Actor',
  future: 'FutureTense',
  pluperfect: 'PluperfectTense',
  perfect: 'PerfectTense',

  PerfectTense: 'PerfectTense',
  PluperfectTense: 'PluperfectTense',
  FutureTense: 'FutureTense',
  PastTense: 'PastTense',
  PresentTense: 'PresentTense',
};

const irregulars = require('./data/irregular_verbs.js');
let verbs = require('./data/verbs.js').concat(Object.keys(irregulars));
for (let i = 0; i < verbs.length; i++) {
  const o = verb_conjugate(verbs[i]);
  Object.keys(o).forEach(function(k) {
    if (k && o[k] && verbMap[k]) {
      lexicon[o[k]] = verbMap[k];
    }
  });
}

let orgs = require('./data/organisations.js');
addArr(orgs.organisations, 'Noun');
addArr(orgs.suffixes, 'Noun');

let places = require('./data/places.js');
addArr(places.countries, 'Place');
addArr(places.cities, 'Place');

require('./data/adjectives.js').forEach(function(s) {
  lexicon[s] = 'Adjective';
  lexicon[to_comparative(s)] = 'Comparative';
  lexicon[to_superlative(s)] = 'Superlative';
});
addObj(require('./data/convertables.js'));

addArr(require('./data/abbreviations.js').abbreviations, 'Abbreviation');
addArr(require('./data/demonyms.js'), 'Adjective');
addArr(require('./data/honourifics.js'), 'Honourific');
addArr(require('./data/uncountables.js'), 'Noun');
addArr(require('./data/dates.js'), 'Date');
addArr(require('./data/numbers.js'), 'Value');
  • ¶

    a little fancy

    addArr(Object.keys(require('./data/firstnames.js')), 'Person');
  • ¶

    add irregular nouns

    const irregNouns = require('./data/irregular_nouns.js');
addArr(fns.pluck(irregNouns, 0), 'Noun');
addArr(fns.pluck(irregNouns, 1), 'Plural');

addObj(require('./data/misc.js'));
addObj(require('./data/multiples.js'));
addObj(require('./data/phrasal_verbs.js'));
  • ¶

    just in case

    delete lexicon[false];
delete lexicon[true];
delete lexicon[undefined];
delete lexicon[null];
delete lexicon[''];
  • ¶

    use ‘Noun’, not ‘NN’

    Object.keys(lexicon).forEach(function(k) {
  if (!grand_mapping[lexicon[k]]) {
  • ¶

    console.log(lexicon[k]);

      }
  lexicon[k] = grand_mapping[lexicon[k]] || lexicon[k];
});
  • ¶

    console.log(Object.keys(lexicon).length) console.log(lexicon)

  • ¶

    console.log(lexicon[‘once again’] === ‘RB’); console.log(lexicon[‘seven’] === ‘Value’); console.log(lexicon[‘sleep’] === ‘VBP’); console.log(lexicon[‘slept’] === ‘VBD’); console.log(lexicon[‘sleeping’] === ‘VBG’); console.log(lexicon[‘canadian’] === ‘JJ’); console.log(lexicon[‘july’] === ‘Value’); console.log(lexicon[null] === undefined); console.log(lexicon[‘dr’] === ‘NNAB’); console.log(lexicon[‘sounds’] === ‘VBZ’); console.log(lexicon[‘look after’] === ‘VBP’); console.log(lexicon[‘tony’] === ‘Noun’); console.log(lexicon[‘loaf’] === ‘Noun’); console.log(lexicon[‘loaves’] === ‘NNS’); console.log(lexicon[‘he’] === ‘PRP’); console.log(lexicon[‘canada’] === ‘Noun’); console.log(lexicon[‘is’]);

    module.exports = lexicon;