TDA opens free “Corpora-for-MT” service to the public

Today, TAUS Data Association (TDA), a non-profit web-based platform for sharing language data, has made translation memory (TM) upload and download open for public users. Anyone in the world can download TMs at no cost as long as they also upload TMs. A user who wants to download more data than they have uploaded will be able to buy credit bundles. This new free public service will drive the adoption and improvement of machine translation (MT) technology. The TDA repository already contains 3.4 billion words in 320 language pairs of human-quality translations, an invaluable resource that has been in use by TDA members for over two years. By opening the platform to public users TDA aims to increase the volume of translated words in more languages and more domains, in turn creating greater benefits for members and new users.

TDA also aims to drive the global translation industry towards greater interoperability. According to a TAUS survey this past month, 38% of the market think that the lack of interoperability costs more than 10% of their translation budget. Forty-three percent don’t know exactly how much it costs them and 12.5% think it costs them more than 30%.

“Sharing translated words not only helps to improve the performance of MT engines, it’s also the best way to drive compliance with standards such as TMX and XLIFF and ensure that tools and vendors can work seamlessly together,” says Jaap van der Meer, director of TDA. “TDA is instrumental in driving innovation and advancement not only of MT technology but also of computer-aided translation, terminology and quality evaluation tools and processes.”

TDA was established in 2008 by 45 founding members. Among the founding members are large IT companies (Microsoft, Oracle, Intel, Adobe, Cisco, EMC) as well as large and small providers of translation services and technologies. TDA currently counts 85 members. Membership is open to freelance translators, universities, government institutions and companies of all sizes. Intellectual property rights to translation memories are protected by universally agreed upon TM Upload and Download Conditions.