The European Commission published an EUR 5.8m call for tender on March 30, 2016 for “tools and resources for CEF automated translation.” It is a follow-up on a previous tender, awarded to a consortium headed by the German Research Center for Artificial Intelligence (DFKI GmbH) in partnership with ELDA, ILSP, and TILDE.
“Automated Translation” is a-so called DSI project, funded to the tune of EUR 8m. DSI, in turn, stands for “digital service infrastructure,” which according to a 55-page background document on the broader Connecting Europe Facility (CEF) initiative are “essential elements of a well-functioning Digital Single Market (DSM).”
Other projects under the same “work program” are e-identification, e-delivery, e-invoicing, cyber security, e-health, and a few others.
Standalone machine translation service
“Automated Translation” is a standalone machine translation service that piggybacks on the EU’s existing MT@EC machine translation system, but extends and adapts it. The main purpose is “to provide multilingual support to other pan-European DSIs so that citizens, administrations and companies in all EU countries can use them in their own language and…offer automated translation services to public administrations to facilitate cross-border information exchange.”
“The automated translation facility will bring significant cost savings and/or increased capacity to the EU institutions’ translation services, currently operating on a budget of roughly EUR 1 billion a year.” – Work Programme 2015
Initially, the CEF program will fund the development and ramp-up of the project; but Automated Translation is meant to eventually become self-funding: “A robust and realistic business model will be developed in order to phase out reliance on CEF funding by 2020.”
And here’s how: “The Automated Translation service can charge the users for automated translations, either based on actual translated volume or by a flat-rate subscription fee.” Users, presumably, are “citizens, administrations and companies” in the EU.
In addition to the carrot of eventual self-financing, the project’s backers justify the investment by promising to bring “significant cost savings” to the European Union’s giant EUR 1bn translation bureaucracy. Very bold.
What’s the EUR 5.8m for?
Lot 1, valued at EUR 1.4m, involves the contractor “assisting” the European Commission in “selecting tools and services,” and shortlisting and testing likely MT systems, which may eventually be purchased.
Lot 2, valued at EUR 2.8m, involves building a language resource repository in all official languages of the EU and CEF-associated countries and providing IPR (intellectual property rights) support and a clearance helpdesk. The Lot 2 contractor is also expected to complement the previous tender won by the DFKI consortium, mentioned above.
In Lot 3, valued at EUR 1.6m, the contract identifies, buys, and anonymises third-party data sets and parallel corpora to train and refine the machine translation engine.
This tender should be interesting for any organization with deep expertise in machine translation technology and related consulting and implementation services―although potential bidders will likely face an uphill battle against the consortium that won the first tender.