This package is partly based on the Snowball stemming algorythm and the javascript adaptation by Kasun Gajasinghe, University of Moratuwa
This package offers 4 NLP tools in javascript for french language :
- Tokenizing
- POS Tagging
- Lemmatizing
- Stemming
npm install nlp-js-tools-french
var NlpjsTFr = require('nlp-js-tools-french');
Corpus to use
var corpus = "Elle semble se nourrir essentiellement de plancton, et de hotdog.";
Configs
var config = {
tagTypes: ['art', 'ver', 'nom'],
strictness: false,
minimumLength: 3,
debug: true
};
New instance with specific corpus and configs
var nlpToolsFr = new NlpjsTFr(corpus, config);
These are the available methods, self-explanatory. Note: The sentence that is passed into the class earlier is automaticaly tokenized.
var tokenizedWords = nlpToolsFr.tokenized;
var posTaggedWords = nlpToolsFr.posTagger();
var lemmatizedWords = nlpToolsFr.lemmatizer();
var stemmedWords = nlpToolsFr.stemmer();
var stemmedWord = nlpToolsFr.wordStemmer("aléatoirement");
Shows config
["semble", "nourrir", "de"]
[{
"id": 1,
"word": "semble",
"pos": [
"VER",
"VER"
]
},
{
"id": 2,
"word": "nourrir",
"pos": [
"VER"
]
},
{
"id": 3,
"word": "de",
"pos": [
"NOM",
"ART:def",
"PRE"
]
}]
[{
"id": 1,
"word": "semble",
"lemma": "sembler"
},
{
"id": 2,
"word": "nourrir",
"lemma": "nourrir"
},
{
"id": 3,
"word": "de",
"lemma": "de"
}]
[{
"id": 1,
"word": "semble",
"stem": "sembl"
},
{
"id": 3,
"word": "nourrir",
"stem": "nourr"
},
{
"id": 5,
"word": "de",
"stem": "de"
}]
{
word: "aléatoirement",
stem: "aléatoir"
}
| Option | Type | Default | Description |
|---|---|---|---|
| tagTypes | Array | ["adj", "adv", "art", "con", "nom", "ono", "pre", "ver", "pro"] |
List of dictionnaries the package will look in, in case you only need verbs or nouns, both or whatever else. If a word does not belong to any type, it is tagged as "UNK". |
| strictness | Bool | false |
If you set the strictness to true and try to POS Tag the word generalement, it will fail because the word is missine its accents. On the other hand, trying to POS Tag the word dé with the strictness set to false well return the types art, pre and nom because the word will match de in these dictionnaries. |
| minimumLength | Int | 1 | Algorythms will ignore words that are shorter than this parameter. |
| debug | Bool | false | Enable console debug |
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent