Neurotõlge is open-source demo project for text machine translation done in the Natural Language Processing research group in the University of Tartu. The project provides convenient methods of data collection, ability to comare different translators and ability to use it as a regular text translator.
Currently, neurotõlge is integrated or have code which simplify integrations with Google, Microsoft, Tilde and UT.
Project is a branch of KAMA.
For machine translation researchers and developers around the world.
Backend was written in Python2 with Flask micro-framework. All missed parts were additionally implemented. (e.g. Multithreading calls to different translators)
Frontend was written in JavaScript, mainly based on ECMAScript5 standard. Additionally, jQuery and AJAX were used. For a beatiful layout, buttons and webpage components Bootstrap3 was in use with a lot of tiny customizations.
You can use neurotõlge with any device, laptop or mobile where Python can be installed and device has a browser.
Installation of Python Beginner's guide.
The next step is creating virtual environment. We recommend to use quite extensive article Virtual environment installation tutorial
When you setup an environment, please, check that you have git installed in case you are planning to contribute to the project or make derivatives from it. Git installation guide
Clone the project to the convenient repository on your device by doing
git clone git@github.com:TartuNLP/neurotolge.git
The last step is installation requirements:
pip install requirements.txt
which can be found in the project.
After all these steps you are ready to work with the project.
If you don't have any preferences regarding environment, we recommend to use PyCharm.
It is the same name of the project. neurotõlge is Estonian word which means Neural Mahine Translation, while neuralmt is English abbreviation with the same sense.
Almost all text translation APIs are paid. Therefore, since we save our resources and unfortunately cannot share security keys with everyone we keep them private. In the meantime, there several translation APIs which have limited freedomness which are not yet implemented. Contributors of the project will be more than happy to add new integrations just let us know through issue tracking page. The same page will be used both for feature queries and issues tracking.