language-support.md

Language Support

Supported languages

Native Support means that the tokenizer and stemmer are included in javascript in NLP.js. BERT Support means that the tokenizer and stemmer are supported through a BERT API made in python. You can see how to create this API here: https://github.com/axa-group/nlp.js/tree/master/examples/80-bert-server

Microsoft Builtins mean that the Builtin Entity extraction is supported directly in javascript, while the ones supported by Duckling requires the deployment of a Duckling instance.

Languages not included in this list can be still supported, but without stemming, only tokenizing. That means less precission, but most of the times can be good enough, as an example you can use it for fantasy languages (at unit tests you'll find tests in klingon from Star Trek).

Locale	Language	Native Support	BERT Support	Microsoft Builtins	Duckling Builtins	Sentiment
af	Afrikaans		X		X
sq	Albanian		X
ar	Arabic	X	X		X	X
an	Aragonese		X
hy	Armenian	X	X			X
ast	Asturian		X
az	Azerbaijani		X
ba	Bashkir		X
eu	Basque	X	X			X
bar	Bavarian		X
be	Belarusian		X
bn	Bengali	X	X		X	X
bpy	Bishnupriya Manipuri		X
bs	Bosnian		X
br	Breton		X
bg	Bulgarian		X		X
my	Burmese		X		X
ca	Catalan	X	X			X
ceb	Cebuano		X
ce	Chechen		X
zh	Chinese (Simplified)	X	X	X	X
zh	Chinese (Traditional)	X	X	X	X
cv	Chuvash		X
hr	Croatian		X		X
cs	Czech	X	X			X
da	Danish	X	X		X	X
nl	Dutch	X	X		X	X
en	English	X	X	X	X	X
et	Estonian		X		X
fi	Finnish	X	X		X	X
fr	French	X	X	X	X	X
gl	Galician	X	X			X
ka	Georgian		X		X
de	German	X	X		X	X
el	Greek	X	X		X	X
gu	Gujarati		X
ht	Haitian		X
he	Hebrew		X		X
hi	Hindi	X	X		X	X
hu	Hungarian	X	X		X	X
is	Icelandic		X		X
io	Ido		X
id	Indonesian	X	X		X	X
ga	Irish	X	X		X	X
it	Italian	X	X		X	X
ja	Japanese	X	X	X	X
jv	Javanese		X
kn	Kannada		X		X
kk	Kazakh		X
ky	Kirghiz		X
ko	Korean	X	X		X	X
la	Latin		X
lv	Latvian		X
lt	Lithuanian	X	X			X
lmo	Lombard		X
nds	Low Saxon		X
lb	Luxembourgish		X
mk	Macedonian		X
mg	Malagasy		X
ms	Malay	X	X
ml	Malayalam		X		X
mr	Marathi		X
min	Minangkabau		X
mn	Mongolian		X		X
ne	Nepali	X	X		X	X
new	Newar		X
nb	Norwegian (Bokmål)	X	X		X	X
nn	Norwegian (Nynorsk)		X
oc	Occitan		X
fa	Persian (Farsi)	X	X			X
pms	Piedmontese		X
pl	Polish	X	X		X	X
pt	Portuguese	X	X	X	X	X
pa	Punjabi		X
ro	Romanian	X	X		X	X
ru	Russian	X	X		X	X
sco	Scots		X
sr	Serbian	X	X			X
hbs	Serbo-Croatian		X
scn	Sicilian		X
sk	Slovak		X		X
sl	Slovenian	X	X			X
az	South Azerbaijani		X
es	Spanish	X	X	X	X	X
su	Sundanese		X
sw	Swahili		X		X
sv	Swedish	X	X		X	X
tl	Tagalog	X	X			X
tg	Tajik		X
ta	Tamil	X	X		X	X
tt	Tatar		X
te	Telugu		X
th	Thai	X	X		X	X
tr	Turkish	X	X		X	X
uk	Ukrainian	X	X		X	X
ur	Urdu		X
uz	Uzbek		X
vi	Vietnamese		X		X
vo	Volapük		X
war	Waray-Waray		X
cy	Welsh		X
fy	West Frisian		X
pa	Western Punjabi		X
yo	Yoruba		X

Sentiment Analysis

Language	AFINN	Senticon	Pattern
Arabic (ar)	X
Armenian (hy)	X
Basque (eu)		X
Bengali (bn)	X
Catalan (ca)		X
Czech (cs)	X
Danish (da)	X
Dutch (nl)			X
English (en)	X	X	X
Finnish (fi)	X
French (fr)			X
Galician (gl)		X
German (de)		X
Greek (el)	X
Hindi (hi)	X
Hungarian (hu)	X
Indonesian (id)	X
Irish (ga)	X
Italian (it)			X
Korean (ko)	X
Lithuanian (lt)	X
Nepali (ne)	X
Norwegian (no)	X
Persian (Farsi) (fa)	X
Polish (pl)	X
Portuguese (pt)	X
Romanian (ro)	X
Russian (ru)	X
Serbian (sr)	X
Slovenian (sl)	X
Spanish (es)	X	X
Swedish (sv)	X
Tagalog (tl)	X
Tamil (ta)	X
Thai (th)	X
Turkish (tr)	X
Ukrainian (uk)	X

Comparision with other NLP products

Locale	Language	Microsoft LUIS	Google Dialogflow	SAP Conversational AI	Amazon LEX	IBM Watson	NLP.js
af	Afrikaans						X
sq	Albanian						X
ar	Arabic	X		X		X	X
an	Aragonese						X
hy	Armenian						X
ast	Asturian						X
az	Azerbaijani						X
ba	Bashkir						X
eu	Basque						X
bar	Bavarian						X
be	Belarusian						X
bn	Bengali						X
bpy	Bishnupriya Manipuri						X
bs	Bosnian						X
br	Breton						X
bg	Bulgarian						X
my	Burmese						X
ca	Catalan			X			X
ceb	Cebuano						X
ce	Chechen						X
zh	Chinese (Simplified)	X	X	X		X	X
zh	Chinese (Traditional)	X	X	X		X	X
cv	Chuvash						X
hr	Croatian						X
cs	Czech					X	X
da	Danish		X	X			X
nl	Dutch	X	X	X		X	X
en	English	X	X	X	X	X	X
et	Estonian						X
fi	Finnish			X			X
fr	French	X	X	X		X	X
gl	Galician						X
ka	Georgian						X
de	German	X	X	X		X	X
el	Greek						X
gu	Gujarati	X					X
ht	Haitian						X
he	Hebrew						X
hi	Hindi	X	X	X			X
hu	Hungarian						X
is	Icelandic						X
io	Ido						X
id	Indonesian		X				X
ga	Irish						X
it	Italian	X	X	X		X	X
ja	Japanese	X	X	X		X	X
jv	Javanese						X
kn	Kannada						X
kk	Kazakh						X
ky	Kirghiz						X
ko	Korean	X	X	X		X	X
la	Latin						X
lv	Latvian						X
lt	Lithuanian						X
lmo	Lombard						X
nds	Low Saxon						X
lb	Luxembourgish						X
mk	Macedonian						X
mg	Malagasy						X
ms	Malay						X
ml	Malayalam						X
mr	Marathi	X					X
min	Minangkabau						X
mn	Mongolian						X
ne	Nepali						X
new	Newar						X
nb	Norwegian (Bokmål)		X	X			X
nn	Norwegian (Nynorsk)						X
oc	Occitan						X
fa	Persian (Farsi)						X
pms	Piedmontese						X
pl	Polish		X	X			X
pt	Portuguese	X	X	X		X	X
pa	Punjabi						X
ro	Romanian						X
ru	Russian		X	X			X
sco	Scots						X
sr	Serbian						X
hbs	Serbo-Croatian						X
scn	Sicilian						X
sk	Slovak						X
sl	Slovenian						X
az	South Azerbaijani						X
es	Spanish	X	X	X		X	X
su	Sundanese						X
sw	Swahili						X
sv	Swedish		X	X			X
tl	Tagalog						X
tg	Tajik						X
ta	Tamil	X					X
tt	Tatar						X
te	Telugu	X					X
th	Thai		X				X
tr	Turkish	X	X				X
uk	Ukrainian		X				X
ur	Urdu						X
uz	Uzbek						X
vi	Vietnamese						X
vo	Volapük						X
war	Waray-Waray						X
cy	Welsh						X
fy	West Frisian						X
pa	Western Punjabi						X
yo	Yoruba						X

Example with several languages

Example with three languages, where one of the language is klingon, to show that NLP will work even with support of the language, because it will use tokenizer but not stemmers.

const { NlpManager } = require('../packages/node-nlp/src');

(async () => {
  const manager = new NlpManager({ languages: ['en', 'ko', 'kl'] });
  // Gives a name for the fantasy language
  manager.describeLanguage('kl', 'Klingon');
  // Train Klingon
  manager.addDocument('kl', 'nuqneH', 'hello');
  manager.addDocument('kl', 'maj po', 'hello');
  manager.addDocument('kl', 'maj choS', 'hello');
  manager.addDocument('kl', 'maj ram', 'hello');
  manager.addDocument('kl', `nuqDaq ghaH ngaQHa'moHwI'mey?`, 'keys');
  manager.addDocument('kl', `ngaQHa'moHwI'mey lujta' jIH`, 'keys');
  // Train Korean
  manager.addDocument('ko', '여보세요', 'greetings.hello');
  manager.addDocument('ko', '안녕하세요!', 'greetings.hello');
  manager.addDocument('ko', '여보!', 'greetings.hello');
  manager.addDocument('ko', '어이!', 'greetings.hello');
  manager.addDocument('ko', '좋은 아침', 'greetings.hello');
  manager.addDocument('ko', '안녕히 주무세요', 'greetings.hello');
  manager.addDocument('ko', '안녕', 'greetings.bye');
  manager.addDocument('ko', '친 공이 타자', 'greetings.bye');
  manager.addDocument('ko', '상대가 없어 남는 사람', 'greetings.bye');
  manager.addDocument('ko', '지엽적인 것', 'greetings.bye');
  manager.addDocument('en', 'goodbye for now', 'greetings.bye');
  manager.addDocument('en', 'bye bye take care', 'greetings.bye');
  manager.addDocument('en', 'okay see you later', 'greetings.bye');
  manager.addDocument('en', 'bye for now', 'greetings.bye');
  manager.addDocument('en', 'i must go', 'greetings.bye');
  manager.addDocument('en', 'hello', 'greetings.hello');
  manager.addDocument('en', 'hi', 'greetings.hello');
  manager.addDocument('en', 'howdy', 'greetings.hello');

  // Train also the NLG
  manager.addAnswer('en', 'greetings.bye', 'Till next time');
  manager.addAnswer('en', 'greetings.bye', 'see you soon!');
  manager.addAnswer('en', 'greetings.hello', 'Hey there!');
  manager.addAnswer('en', 'greetings.hello', 'Greetings!');

  // Train and save the model.
  await manager.train();
  manager.save();

  // English and Korean can be automatically detected
  manager.process('I have to go').then(console.log);
  manager.process('상대가 없어 남는 편').then(console.log);
  // For Klingon, as it cannot be automatically detected, 
  // you must provide the locale
  manager.process('kl', `ngaQHa'moHwI'mey nIH vay'`).then(console.log);
})();