CEF TELECOM GRANT BENEFICIARY
eTranslation and eDelivery used to build new-age machine translation router
Pan-European consortium builds a dynamic router, able to smoothly switch between domain-specific translation engines, to produce the highest possible quality machine translations.
@Photo by Pangeanic
Manuel Herranz, CEO of Pangeanic, at a post-project event in Madrid, Spain, on 12 March 2019, after successful completion of iADAATPA.
- Countries: Spain, Latvia, Lithuania, Ireland
- Organisation: iADAATPA consortium (now called MT-Hub)
- Project: iADAATPA
- Challenge: How to provide high-quality machine translations when a single document can include content from different subject areas and domains?
- Solution: Development of a translation engine router with language and domain detection mechanisms to dynamically switch between machine translation engines whenever a new domain is found
- Building Blocks: eTranslation, eDelivery
- EU funded: 75 %
Cross-border communication is always a challenge. Most European countries have machine translation (MT) systems in place, but many of them are not up to par. For example, Spain has an old system that translated the name of the Ministry Dolores del Campo into “it is pain of field”. The mistake went unnoticed and was published in November, 2018, in a press release by the Ministry of Industry, Commerce and Tourism, causing global amusement1. It was high time to replace old systems with 21st century technology.
To improve MT quality, Pangeanic had the novel idea of building a scalable and secure platform for the provision of multi-vendor automated translation services. The resulting platform, called iADAATPA, is a translation engine router with language and domain detection mechanisms. In other words, it detects the language and domain of a given document and routes it to the translation engine best suited to translate content. This frees public administrations of vendor lock-in, allowing them to contract multiple vendors for MT services.
The platform is even capable of switching between MT engines within a document and for different parts of the document, as soon as it detects a change in language or domain. To develop iADAATPA, a consortium was founded with Commission funding. The project started in autumn 2017 and ran for 1.5 years.
Who was involved?
The iADAATPA consortium was made up of Pangeanic (Spain), SEAD - the Secretary of State for Digital Advancement (Spain), Everis (Spain), Prompsit (Spain), SEGGITUR (Spain), Tilde (Latvia), KantanMT (Ireland) and the Dublin City University (Ireland).
The consortium got to work and chose standards and technologies promoted through the European Commission's Connecting Europe Facility (CEF) programme, because they have been proven in practice as part of Europe's interoperable digital ecosystem and are hence low in risk. In more detail, the iADAATPA uses CEF's building blocks eTranslation and eDelivery. Building blocks are the European Commission's freely available standards-based digital solutions to help European public and private sector organisations develop digital services faster and more efficiently. eTranslation provides translation engines in seven EU domains and a web service to facilitate cross-border communication in all official EU languages, Icelandic and Norwegian. eDelivery provides the technical means to exchange digital data and documents securely and reliably.
An integration was made from iADAATPA to eTranslation, which provided valuable language resources and MT engines in seven EU domains. eDelivery (AS4 Domibus) was used to ensure the secure and reliable exchange of the documents to be translated. To further complement the capabilities of eTranslation, consortium members made additional integrations to their own MT engines to cover domains not provided by eTranslation.
iADAATPA makes use of neural networks algorithms based on artificial intelligence, instead of the traditional statistical MT methods based word sequence patterns, for example. Consortium partners and the European Commission agreed to go for the neural-networks based approach early on in the project. The platform was implemented and tested in the following use cases:
- Generalitat Valenciana, Spain - The Open Data portal of the regional government's transparency department
- SEGITTUR Spain - Digital services provided by the agency for innovation and tourism technology based in Madrid
- Lithuanian Parliament - Translation of parliamentary proceedings in Lithuanian Parliament
- Dublin City College website
- Deutsche Welle
Results and benefits
The iADAATPA platform was proven ready for consumer use based on an assessment of its language and domain detection abilities. iADAATPA will change the MT landscape by offering a vendor and technology neutral platform that will increase MT efficiency and competition and eventually, lead to more competitive translation prices on the market. By enhancing access to different MT providers, it will also help realise the Digital Single Market in Europe.
The translation quality of each individual MT engine was also tested in their respective domains with positive results. For example, the project's final report stated that Pangeanic's MT system, specialised in tourism and PA OpenData, was "ranked as ‘best’ more than 67 % of the times, against 7.32 % of the times for the baseline", with the baseline being another leading MT engine in the market. This shows that highly specialised translation engines are a must for achieving top-quality translations, and iADAATPA is needed for smoothly switching between domains and engines. Based on the positive outcome of the project, the Spanish Administration has decided to adopt iADAATPA, now called MT-hub platform, as a core component of their national translation infrastructure.
Results also surpassed all expectations of the Lithuanian Parliament. Lithuania had had bad experiences with MT technologies before, as their language is hard to translate with traditional statistical translation methods. With iADAATPA, productivity and MT quality significantly increased. The consortium reported productivity increases of around 40 % in achieving final human quality translations.
All members of the consortium had complementary skills and assets, which made collaboration very fruitful. Professor Andy Way from the Dublin City University rejoiced that "the iADAATPA consortium partners proved such a high level of professionalism that I'd love to work with all of them again. This is the proof of a well-run consortium."
The consortium is now focusing on an extension of the project to get the word out to public administrations on this new service, and to help public administrations adopt a framework to manage MT services at different levels.
What to know more?
Visit us to learn more about the Building Blocks behind this solution. If you are interested in using our Building Blocks for a solution of your own, we would be happy to help you get started. Let’s connect Europe, together.
Fiorentino, Michael-Ross. 'Pain of the field': the embarrassing automatic translation of the Spanish Ministry of Industry. Euronews, 19 Nov. 2018, https://www.euronews.com/2018/11/19/pain-of-the-field-the-embarrassing-automatic-translation-of-the-spanish-ministry-of-indus. Accessed 8 Mar. 2019.
This page has no comments.