Project

EuroMatrixPlus

EuroMatrixPlus

EuroMatrixPlus

  • Duration:

Europe faces a growing economic and societal challenge due to its vast diversity of languages, and machine translation technology holds promise as a means to address this challenge. The goals of EuroMatrixPlus are:

Firstly, the project will continue the rapid advance of machine translation technology, creating example systems for every official EU language, and providing other machine translation developers with our infrastructure for building statistical translation models.

Secondly, the project will continue and broaden the controlled systematic investigation of different approaches and techniques to accelerate the scientific evolution of novel methods, including both selection and cross-fertilization. The aim is to arrive at scientifically well understood novel combinations of methods that are proven superior to the state of the art.

Thirdly, the project will focus on bringing machine translation to the users, in addition to focusing on scientific advances. Because our statistical models are derived from example translations, we believe that there is potential for a synergistic relationship in which users suggest improvements to the system by post-editing its output, and the system improves itself by learning from user feedback.

Fourthly, the project will contribute to the growth and competitiveness of the European MT research scene and infrastructure through its open international competitive shared tasks and living community supported surveys of resources, tools, systems and their respective capabilities.

In bringing MT to the users, EuroMatrixPlus focuses on two different types of users: (a) professional translators and translation agencies working for private corporations, administrations, and other organisations, and (b) lay users who create content on a volunteer basis by translating foreign materials into their own languages. The project will investigate how these users can benefit from state of the art machine translation, and conversely, how machine translation can benefit from user corrections.

EuroMatrixPlus will create an openly accessible sample application that enables users to automatically translate news stories and web pages from any European language into any other, and whose corrections will be exploited as data for improving translation technology.

Partners

Publications about the project

Rui Wang, Petya Osenova, Kiril Simov

In: Proceedings of the Sixth Workshop on Syntax, Semantics and Structure in Statistical Translation. Workshop on Syntax, Semantics and Structure in Statistical Translation (SSST-6) July 12 Jeju South Korea Pages 10-19 Association for Computational Linguistics 7/2012.

To the publication
Christian Federmann, Dagmar Gromann, Thierry Declerck, Sabine Hunsicker, Hans-Ulrich Krieger, Gerhard Budin

In: Guadalupe Aguado de Cea, Mari Carmen Suárez-Figueroa, Raúl García-Castro, Elena Montiel-Ponsoda (editor). Proceedings of the 10th Terminology and Knowledge Engineering Conference. Terminology and Knowledge Engineering Conference (TKE-2012) New frontiers in the constructive symbiosis of terminology and knowledge engineering June 20-21 Madrid Spain Pages 166-175 TKE Madrid 6/2012.

To the publication
Sabine Hunsicker, Yu Chen, Christian Federmann

In: Proceedings of the Seventh Workshop on Statistical Machine Translation. Workshop on Statistical Machine Translation (WMT-12) Montréal Québec Canada Pages 312-316 Association for Computational Linguistics 6/2012.

To the publication

German Research Center for Artificial Intelligence
Deutsches Forschungszentrum für Künstliche Intelligenz