"MT for Everyone" is a tutorial series that aims to engage translators and translation agencies in the customization of machine translation systems. It is an opportunity that would particularly benefit those senior translators interested in availing themselves of their own translation memories and corpora. By integrating the latest technology in custom MT and post-editing into the pipeline, the translation cost could significantly drop as productivity levels rise (Note that custom MT results are domain dependent).
Before the revolutionary breakthrough of deep learning, the translator mostly stayed out of the development of machine translation systems, giving way to the circulation of many myths, like it represented an economic threat to translators or a loss of respect to their cultural, linguistic skills and knowledge. Nonetheless, it is now evident that
1) translations of human quality are impossible to achieve without a human translator;
2) the danger of extinction against the profession is to be discarded from the moment in which the technology with the most potential is the one that depends on human translations.
In addition to this, one cannot refuse the fact that machine translation belongs to the list of everyday tools of translators, just like translation memories or Computer Aided Translation (CAT) software. Considering the latest trends in translation technologies, the increasing interest in the customization of machine translation engines as an added service is reflected on the greater offer of software dedicated to this practice. Limitations such as the lack of a graphical interface in software, like Moses, or the lack of knowledge base on how to use the terminal and other cumbersome processes that hindered the translator's agency have been overcome. Today, we count with SaaS platforms which feature a user-friendly interface, instant automated metric reports, which allow a fast appraisal of the engine's quality and a variety of tools to refine the engine's translation proficiency. Along with this, open data repositories such as the Opus Corpus make most linguistic resources necessary to train an engine available to anyone. The availability of all of these resources shows that corpus-driven approaches are more accessible than what we think. These series of events that are currently stirring so much speculation about the future of translators in regard to this technology was Qabiria's motivation to produce "MT for Everyone."
What is "MT for Everyone"?
"MT for Everyone" consists of 15 episodes of less than 3 minutes long each. Throughout this series, we attempt to cover all phases related to the customization of an engine and the most common data-driven approaches. You will learn how to obtain a corpus, prepare and clean your training data sets, appraise and refine your engine's output quality. This series targets both, newcomers to the translation industry and senior translators, who would like to take advantage of this technology to boost up their productivity.
When can we expect it?
The official release date of "MT for Everyone" is on August 1st, 2018. From then on, viewers can expect to find a new episode on our YouTube channel every Wednesday. You may also want to subscribe to our channel to get notified about our latest program updates.
By now, we leave you with the introductory video of "MT for Everyone," which should give you a hint of what is coming in the next episodes.
Follow our latest video uploads on our YouTube channel.