SYSTRAN Enhances Automated Professional Translation by Launching the First Catalog of Industry-Specific Translation Models
New tools and models, created to meet the level of quality expected by professional users, are available in self-service in the cloud and covers industries such as agriculture, education, finance, IT, legal and health
SYSTRAN, the leader in neural machine translation technology, announces the launch of its new multilingual catalog of industry-specific translation models. These unique models are designed by a community of experts called ‘trainers’—owners, translators, linguists, universities, public organizations, etc.—and offered in self-service to professional users—SMBs, translators, business owners—facing the limitation of generic translation.
The objective of SYSTRAN Marketplace is to meet the level of quality expected by professionals who need a translation customizable to their business jargon, as well as industry-specific vocabulary and brand voice. The new technology will cover industries such as agriculture, education, finance, IT, legal and health.
Innovation is in the ecosystem
“We have been committed to the same goal for 50 years,” said CEO of SYSTRAN, Jean Senellart. “We believe that providing quality and customized translation is possible thanks to the added value of humans, rather than the technology itself. Our neural technology and training engines may be a highly innovative ‘toolbox’, but our success will depend on our ability to convince the world’s largest community of experts to join us and use our platform to create, train, and offer the widest catalog of languages and models.”
Last year, TED joined the community and granted SYSTRAN, for the first time in its history, the right to use the bilingual data from its prestigious conferences to create translation models in 10 languages, highly adapted to educational, scientific and technical content. Independent human evaluations have shown impressive improvements, both in precision and fluidity, compared to generic models. For 41 percent of sentences in the domain, these models even exceed the reference human translations in quality.
“The fact that our inaugural collaboration in the AI space is focused on neural machine translation models built from translations of TED Talks in multiple languages feels natural,” said Alex Hofmann, Director, Global Distribution & Licensing at TED. “These are now available on a licensed basis to help enterprises and organizations meet their most sophisticated translation needs.”