langdetect-ensemble

Detect the language of text, support ensembling of multiple models through fastText. langdetect-ensemble uses Franc by default in case no language models are loaded

Install

npm:

npm install langdetect-ensemble

Usage

Loading the library:

var fs = require('fs'); // Use to automatically load all models from given folder
var lang = require('langdetect-ensemble')

Initializing LangDetect object:

var dir_models = __dirname + '/path/to/your/model/folder/'
langDet = new lang.langDetect()

fs.readdirSync(dir_models).forEach(function(file) {
      let model_file = dir_models +'/'+file;
      try {
        langDet.addModel(model_file)
      } catch (error) {
        console.log(error);
      }
  });

Setting weights of model:

langDet.addModel('./lid.176.ftz')
langDet.setWeights([0.3, 0.7]) // Here, a vote by franc counts for 0.3, and a vote by the FastText model counts for 0.7

Using LangDetect object for language detection:

langDet.predict('chào buổi sáng').then((lang) => {
    console.log(lang);
    // expected output: 'vie'
});

License

MIT © [Cuong V Nguyen]