GitHub - saffsd/langid.js: An off-the-shelf client-side language identification module for JavaScript. (original) (raw)

Introduction

langid.js is a direct port of the language identifier implemented by langid.py. The theory behind the method is described in two published research papers [1,2]. langid.js does not implement the training of the model, and instead provides a tool ldpy2ldjs.py to convert models trained with the langid.py training tools.

Demonstration

Open demo.html in a browser. langid.js uses TypedArrays so a browser that supports them is required.

Usage

The models and the actual classifier are distributed as two separate javascript files, and both must be included in a page for the langid.js to work. In this repository, I initially provide langid-model-acquis.js, a toy 4-language model based on only JRC-Acquis data, useful for testing and development purposes, as well as langid-model-full.js, the same model that is packaged by default with langid.py.

References

[1] http://aclweb.org/anthology-new/I/I11/I11-1062.pdf [2] http://www.aclweb.org/anthology/P/P12/P12-3005.pdf