AMTA 2018 | Tutorial | ModernMT: Open-Source Adaptive Neural MT for Enterprises and Translators
Nowadays, computer-assisted translation (CAT) tools represent the dominant technology in the translation market – and those including machine translation (MT) engines are on the increase. In this new scenario, where MT and post-editing are becoming the standard portfolio for professional translators, it is of the utmost importance that MT systems are specifically tailored to translators.
In this tutorial, we will present ModernMT, a new open-source MT software whose development was funded by the European Union. ModernMT targets two use cases: enterprises that need dedicated MT services; and professional translators working with CAT tools. This tutorial will focus on both use cases.
In the first part, we will present the ModernMT open source software architecture and guide the audience through its installation on an AWS instance. Then, we demonstrate how to create a new adaptive Neural MT engine from scratch, how to feed its internal memory, and finally how to query it.
In the second part, we will introduce ModernMT’s most distinguishing features when used through a CAT tool: (i) ModernMT does not require any initial training: as soon as translators upload their translation memories in the CAT tool, ModernMT seamlessly and quickly learns from this data; (ii) ModernMT adapts to the content to be translated in real time: the system leverages the training data most similar to the document being translated; (iii) ModernMT learns from user corrections: during the translation workflow, ModernMT constantly learns from the post-edited sentences to improve its translation suggestions. In particular, we will demonstrate ModernMT within MateCat, a popular online professional CAT tool.
In this tutorial, participants will learn about industry trends aiming to develop MT focusing on the specific needs of enterprises and translators. They will see how current state-of-the-art MT technology is being consolidated into a single, easy-to-use product capable of learning from – and evolving through – interaction with users, with the final aim of increasing MT-output utility for the translator in a real professional environment.
Marcello Federico (MMT, FBK) and Davide Caroselli (MMT)
MT users, specialists, integrators, developers, managers, decision makers.
Want to read more about the conference?