Skip to content

Nodejs binding for fasttext representation and classification.

License

Notifications You must be signed in to change notification settings

rse/FOREIGN-node-fasttext

 
 

Repository files navigation

node-fasttext

Nodejs binding for fasttext representation and classification.

MIT License npm version downloads Travis Appveyor

This is a link to the Facebook fastText. A Library for efficient text classification and representation learning.

  • FASTTEXT_VERSION = 12;
  • FASTTEXT_FILEFORMAT_MAGIC_INT32 = 793712314;

Installation

Using npm:

npm install fasttext --save

fastText Classifier

According to fasttext.cc. We have a simple classifier for executing prediction models about cooking from stackexchange questions:

const path = require('path');
const fastText = require('fasttext');

const model = path.resolve(__dirname, './model_cooking.bin');
const classifier = new fastText.Classifier(model);

classifier.predict('Why not put knives in the dishwasher?', 5)
    .then((res) => {
        if (res.length > 0) {
            let tag = res[0].label; // __label__knives
            let confidence = res[0].value // 0.8787146210670471
            console.log('classify', tag, confidence, res);
        } else {
            console.log('No matches');
        }
    });

The model haved trained before with the followings params:

const path = require('path');
const fastText = require('fasttext');

let data = path.resolve(path.join(__dirname, '../data/cooking.train.txt'));
let model = path.resolve(path.join(__dirname, '../data/cooking.model'));

let classifier = new fastText.Classifier();
let options = {
    input: data,
    output: model,
    loss: "softmax",
    dim: 200,
    bucket: 2000000
}

classifier.train('supervised', options)
    .then((res) => {
        console.log('model info after training:', res)
        // Input  <<<<< C:\projects\node-fasttext\test\data\cooking.train.txt
        // Output >>>>> C:\projects\node-fasttext\test\data\cooking.model.bin
        // Output >>>>> C:\projects\node-fasttext\test\data\cooking.model.vec
    });

Or you can train directly from the command line with fasttext builded from official source:

# Training
~/fastText/data$ ./fasttext supervised -input cooking.train -output model_cooking -lr 1.0 -epoch 25 -wordNgrams 2 -bucket 200000 -dim 50 -loss hs
Read 0M words
Number of words:  8952
Number of labels: 735
Progress: 100.0%  words/sec/thread: 1687554  lr: 0.000000  loss: 5.247591  eta: 0h0m 4m

# Testing
~/fastText/data$ ./fasttext test model_cooking.bin cooking.valid
N       3000
P@1     0.587
R@1     0.254
Number of examples: 3000

Nearest neighbor

Simple class for searching nearest neighbors:

const path = require('path');
const fastText = require('fasttext');

const model = path.resolve(__dirname, './skipgram.bin');
const query = new fastText.Query(model);

query.nn('word', 5, (err, res) => {
    if (err) {
        console.error(err);
    } else if (res.length > 0) {
        let tag = res[0].label; // letter
        let confidence = res[0].value // 0.99992
        console.log('Nearest neighbor', tag, confidence, res);
    } else {
        console.log('No matches');
    }
});

Build from source

See Installation Prerequisites.

# install dependencies and tools
npm install

# build node-fasttext from source
npm run build

# run unit-test
npm test

Contributing

Pull requests and stars are highly welcome.

For bugs and feature requests, please create an issue.

About

Nodejs binding for fasttext representation and classification.

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • C++ 85.5%
  • JavaScript 7.5%
  • Python 7.0%