@nlpjs-neo/lang-it

Part of NLP.js Neo, a refreshed and maintained fork of axa-group/nlp.js. The packages are ESM only, written in TypeScript, and require Node.js 22.12 or later.

TABLE OF CONTENTS

• Installation
• Normalization
• Tokenization
• Identify if a word is an italian stopword
• Remove stopwords from an array of words
• Change the stopwords dictionary
• Stemming word by word
• Stemming an array of words
• Normalizing, Tokenizing and Stemming a sentence
• Remove stopwords when stemming a sentence
• Sentiment Analysis
• Example of usage on a classifier
• Contributing
• Code of Conduct
• Who is behind it
• License

Installation

You can install @nlpjs-neo/lang-it:

    pnpm add @nlpjs-neo/lang-it

Normalization

Normalization of a text converts it to lowercase and remove decorations of characters.

import { NormalizerIt } from '@nlpjs-neo/lang-it';

const normalizer = new NormalizerIt();
const input = 'Questo dòvrebbe essere normalizzato';
const result = normalizer.normalize(input);
console.log(result);
// output: questo dovrebbe essere normalizzato

Tokenization

Tokenization splits a sentence into words.

import { TokenizerIt } from '@nlpjs-neo/lang-it';

const tokenizer = new TokenizerIt();
const input = 'Questo dovrebbe essere tokenizzato';
const result = tokenizer.tokenize(input);
console.log(result);
// output: [ 'Questo', 'dovrebbe', 'essere', 'tokenizzato' ]

Tokenizer can also normalize the sentence before tokenizing, to do that provide a true as second argument to the method tokenize

import { TokenizerIt } from '@nlpjs-neo/lang-it';

const tokenizer = new TokenizerIt();
const input = 'Questo dovrebbe essere tokenizzato';
const result = tokenizer.tokenize(input, true);
console.log(result);
// output: [ 'questo', 'dovrebbe', 'essere', 'tokenizzato' ]

Identify if a word is an italian stopword

Using the class StopwordsIt you can identify if a word is an stopword:

import { StopwordsIt } from '@nlpjs-neo/lang-it';

const stopwords = new StopwordsIt();
console.log(stopwords.isStopword('uno'));
// output: true
console.log(stopwords.isStopword('sviluppatore'));
// output: false

Remove stopwords from an array of words

Using the class StopwordsIt you can remove stopwords form an array of words:

import { StopwordsIt } from '@nlpjs-neo/lang-it';

const stopwords = new StopwordsIt();
console.log(stopwords.removeStopwords(['ho', 'visto', 'uno', 'sviluppatore']));
// output: [ 'visto', 'sviluppatore' ]

Change the stopwords dictionary

Using the class StopwordsIt you can restart it dictionary and build it from another set of words:

import { StopwordsIt } from '@nlpjs-neo/lang-it';

const stopwords = new StopwordsIt();
stopwords.dictionary = {};
stopwords.build(['ho', 'visto']);
console.log(stopwords.removeStopwords(['ho', 'visto', 'uno', 'sviluppatore']));
// output: [ 'uno', 'sviluppatore' ]

Stemming word by word

An stemmer is an algorithm to calculate the stem (root) of a word, removing affixes.

You can stem one word using method stemWord:

import { StemmerIt } from '@nlpjs-neo/lang-it';

const stemmer = new StemmerIt();
const input = 'svilupp';
console.log(stemmer.stemWord(input));
// output: program

Stemming an array of words

You can stem an array of words using method stem:

import { StemmerIt } from '@nlpjs-neo/lang-it';

const stemmer = new StemmerIt();
const input = ['ho', 'visto', 'uno', 'sviluppatore'];
console.log(stemmer.stem(input));
// outuput: [ 'ho', 'vist', 'uno', 'svilupp' ]

Normalizing, Tokenizing and Stemming a sentence

As you can see, stemmer does not do internal normalization, so words with uppercases will remain uppercased. Also, stemmer works with lowercased affixes, so sviluppatore will be stemmed as svilupp but SVILUPPATORE will not be changed.

You can tokenize and stem a sentence, including normalization, with the method tokenizeAndStem:

import { StemmerIt } from '@nlpjs-neo/lang-it';

const stemmer = new StemmerIt();
const input = 'Ho visto uno SVILUPPATORE';
console.log(stemmer.tokenizeAndStem(input));
// output: [ 'ho', 'vist', 'uno', 'svilupp' ]

Remove stopwords when stemming a sentence

When calling tokenizeAndStem method from the class StemmerIt, the second parameter is a boolean to set if the stemmer must keep the stopwords (true) or remove them (false). Before using it, the stopwords instance must be set into the stemmer:

import { StemmerIt, StopwordsIt } from '@nlpjs-neo/lang-it';

const stemmer = new StemmerIt();
stemmer.stopwords = new StopwordsIt();
const input = 'Ho visto uno sviluppatore';
console.log(stemmer.tokenizeAndStem(input, false));
// output: [ 'vist', 'svilupp' ]

Sentiment Analysis

To use sentiment analysis you'll need to create a new Container and use the plugin LangIt, because internally the SentimentAnalyzer class try to retrieve the normalizer, tokenizer, stemmmer and sentiment dictionaries from the container.

import { Container } from '@nlpjs-neo/core';
import { SentimentAnalyzer } from '@nlpjs-neo/sentiment';
import { LangIt } from '@nlpjs-neo/lang-it';

(async () => {
  const container = new Container();
  container.use(LangIt);
  const sentiment = new SentimentAnalyzer({ container });
  const result = await sentiment.process({
    locale: 'it',
    text: 'amore per i gatti',
  });
  console.log(result.sentiment);
})();
// output:
// {
//   score: 0.25,
//   numWords: 4,
//   numHits: 2,
//   average: 0.0625,
//   type: 'pattern',
//   locale: 'it',
//   vote: 'positive'
// }

The output of the sentiment analysis includes:

• score: final score of the sentence.
• numWords: total words of the sentence.
• numHits: total words of the sentence identified as having a sentiment score.
• average: score divided by numWords
• type: type of dictionary used, values can be afinn, senticon or pattern.
• locale: locale of the sentence
• vote: positive if score greater than 0, negative if score lower than 0, neutral if score equals 0.

Example of usage on a classifier

import { containerBootstrap } from '@nlpjs-neo/core';
import { Nlp } from '@nlpjs-neo/nlp';
import { LangIt } from '@nlpjs-neo/lang-it';

(async () => {
  const container = await containerBootstrap();
  container.use(Nlp);
  container.use(LangIt);
  const nlp = container.get('nlp');
  nlp.settings.autoSave = false;
  nlp.addLanguage('it');
  // Adds the utterances and intents for the NLP
  nlp.addDocument('it', 'Addio per ora', 'greetings.bye');
  nlp.addDocument('it', 'arrivederci e stai attento', 'greetings.bye');
  nlp.addDocument('it', 'molto bene a dopo', 'greetings.bye');
  nlp.addDocument('it', 'devo andare', 'greetings.bye');
  nlp.addDocument('it', 'ciao', 'greetings.hello');
  
  // Train also the NLG
  nlp.addAnswer('it', 'greetings.bye', 'fino alla prossima volta');
  nlp.addAnswer('it', 'greetings.bye', 'A presto!');
  nlp.addAnswer('it', 'greetings.hello', 'Ciao, come stai');
  nlp.addAnswer('it', 'greetings.hello', 'Saluti!');
  await nlp.train();
  const response = await nlp.process('it', 'devo andare');
  console.log(response);
})();

Contributing

You can read the guide of how to contribute at Contributing.

Code of Conduct

You can read the Code of Conduct at Code of Conduct.

Who is behind it

NLP.js was created and developed by AXA Group Operations Spain S.A., with Jesus Seijas as its main author; AXA Group is no longer involved in maintaining it. nlpjs-neo is a fork of that project, currently maintained by Igor Savin and the contributors to the fork.

License

Copyright (c) AXA Group Operations Spain S.A.

Copyright (c) 2026 Igor Savin

Released under the MIT License.