With the growth of the Internet, hyper-globalisation and the needs of multilingual markets, the translation services industry is attempting to cover the rising volumes of translation demand. Machine translation provides a technological solution where there is a need for rapid translation into multiple languages and linguistic services providers have been able to exploit the advances in customised machine translation.
Customisation offers are mostly based on approaches to machine translation that rely on statistical methods, which enable the development of translation systems from specific textual material, be it based on public data or data provided by the clients themselves. For closed domains, the translations obtained with customised machine translation are more precise than the ones offered by generic systems. Under this data-driven approach, domain-specific data exploitation requires the use of a a wide range of methods, given the heterogeneous types, volumes and sources of exploitable data, and the need to extract the appropriate information to develop fully customised statistical translation systems.
The main objective of project AdapTA is the development of a solution that enables the generation of customised machine translation systems from heterogeneous sources by exploiting client-specific data as well as online content. The project covers the development and integration of optimal methods for the selection, integration and use of domain specific data within statistical translation models and the precise evaluation of the benefits obtained from customised systems in terms of quality and usability.
This project has received in funding from the Ministerio de Economía y Competitividad of the Spanish Government. This program is funded by EU through the operational programme with the aim of "fostering quality technology development, innovation and research"
01/09/2015 - 31/12/2017