Machine Translation (MT) is being used by many people now as a productivity tool, with demonstrable success. There is already a wide variety of use-cases, but more are emerging where MT is the only solution. Many of these new, emerging use-cases for raw MT and post-edited MT require different levels of human engagement, and different levels of quality. Despite the fact that more and more people are finding MT to be useful, the introduction of MT into the translation pipeline has been less than wholly embraced by the entire translator community. Nonetheless, we argue – as we have done for a number of years now – that the translation community has little to fear from MT, and that where human input is required, the translator remains the most important link in the chain.

In this gentle introduction to machine translation, we will cover the basics of the phrase-based and neural machine translation approaches. In particular, for each approach we will present the underlying models, the overall architecture, the machine learning framework, and the inference methods. We will also discuss and compare the general properties and quality-cost trade offs of each approach, to point out their strengths and weaknesses as well as to suggest under which condition one approach would be preferable over the other. Finally, we will present a few recent advances in neural MT, to witness the relentless research effort going on worldwide in this field.

Demonstration of the key features of the MMT system when used by the translator through the MateCat tool, namely (i) MMT does not require any initial training time: as soon as translators upload their translation memories in the CAT tool, MMT seamlessly and quickly learns from this data; (ii) MMT adapts in real time to the content to be translated: the system leverages those training data that are most similar to the document being translated; (iii) MMT learns from user corrections: during the translation workflow, MMT is constantly learning from the post-edited sentences to improve its translation suggestions.

Typically, MT is used to pre-translate texts that the translator is left with to post-edit, i.e. correct the mistakes that the machine made. Lilt replaces post-editing with interactive and adaptive machine assistance and gives an entirely different experience. Lilt effectively integrates MT, translation memories and termbases into one interface that learns and adjusts from translators or even translation teams in real-time. Interactive MT is more accurate because it can observe what the translator has typed and update the suggestions based on all available information. Adaptive MT saves time by reducing the repetition of mistakes that were already corrected. Lilt's approach leaves translators in control of their workflow and provides a new integrated interface designed to make it easy to use advanced technology.

Evaluation of machine translation is crucial in order to understand the potential as well as the limitations of this technology to be used in the translation workflow. We will present both standard human evaluation methods (e.g. fluency and adequacy judgements) and state-of-the-art automatic evaluation metrics (AEMs), focusing on the role of human translators in evaluating MT output and in interpreting AEMs.

Participants will be presented with results from a quality evaluation of neural vs. phrase-based MT carried out at WeLocalize

In the last decade the explosion of global content in the cloud has accelerated the pace of language technology R&D progress towards total connectivity and automation, rapid processing and high productivity to meet the market needs for 24/7 on-demand services at an unprecedented scale. We are entering the era of total convergence of technology and human talent. What does this mean for the translation industry landscape beyond 2020?

LSP Innovation Stories

The aim of the panel is to share knowledge and experience about growing businesses and current trends in the translation industry. Since the most innovative LSPs are turning to machine translation to increase their productivity and market competitiveness, the panel will discuss issues such as how LSPs see the translator’s profession now and in the future, what skills are and will be highly valued from language professionals, how LSPs are promoting MT technology adoption in the translator community, how translators can best promote their MT and post-editing skills.

Moderator: Olga Blasco
Panelists:Alessandro Cattelan, Manuel Herranz, Tanja Schmidt

MateCat is a free, online translation platform. What makes it unique is how fast you translate with it and how easy it is to use. Nevertheless, there’s a number of features that are worth a more in-depth look. In this session, we’ll learn how to translate a document the fast way, going from project creation to translation in a couple of minutes. After that, we’ll dive into some of the most useful features of the tool, such as translation memory and glossary management, machine translation, automated tag management, dictation, segmentation and QA.

In this session we’ll cover the project management features of the tool. We’ll see how you can split your translation projects in up to 50 jobs for multiple translators to work simultaneously with the same translation memories and glossaries. You’ll discover a new way to measure translation productivity and the quality of the translation memories and machine translation you are using. Finally, we’ll cover project management aspects such as how to create projects, manage them online with your team of project managers, assign projects to your translators, outsource the most difficult languages, keep track of projects’ progress and quality.

This session is all about you. We’ll revisit the key functionalities of MateCat by actually creating a project and translating some files.

In questa sessione capiremo quali sono le migliori prassi per la creazione e la gestione di un progetto.

In questa sessione affronteremo l’ambiente di traduzione vero e proprio con attenzione a quelle che sono le nuove funzionalità della versione SDL Trados Studio 2017.

Prova pratica di utilizzo di SDL Trados Studio 2017