The EuroMatrix Project
EuroMatrix aims at increasing the availability and quality of machine translation (MT) of technical, social, legal and political documents across all pairs of EU languages, including the languages of new member states.
The number and scope of translations in business, administration, politics and culture is increasing every year. The exploding demand for text translations is a direct consequence of open trade practices, globalisation and, in a European context, the integration and enlargement of the EU. The legion of translators is growing. However, with the general trend towards specialisation, the demand on the competences of specialised translators also increases. This leads to a general increase in the costs of high-quality translations.
To reduce translation time and costs, translators have been using specialised computer programs for years, which support the translation process by choosing appropriate suggestions for translations from the body of previous translations. However, these systems cannot, for example, translate sentences which are not available in the program’s memory.
This is only made possible by machine translation systems, which are often considered as being the only realistic solution to managing the communication problems of a multilingual world society. More and more experts consider advances in MT as a critical precondition for the sustainable preservation of the linguistic and cultural diversity on our planet.
Although experts agree among themselves on the immense demand for machine translation, opinions concerning the possibilities of such systems and the quality of their output with regard to complex texts differ widely and it is true that the real breakthrough in machine translation is yet to come.
However, the impressive research advances in the area of statistical machine translation in recent years give rise to hope. In such systems, programs learn translation patterns from large bodies of already-translated text. Their output is still far from perfect, but they are improving every year. As their weak spots are different from those of the conventional, rule-based translation programs designed by linguistics, a solution may be found in a clever combination of the two methods.
This is the aim of the EuroMatrix project which will design and investigate novel combinations of statistical techniques and linguistic knowledge sources as well as hybrid MT architectures in order to combine the accuracy of rule-based techniques with the adaptivity of data-driven approaches. For most translation directions that involve languages of the new and near-term prospective member states, the project will provide baseline MT functionality for the first time.
The ambitious objective of the project is to develop and test translation systems for the 23 official languages of the European Union. It will of course not be possible to focus on each language pair in the same way but one of the practical results of the project will be the continuously updated evaluation of the MT technology available for each language pair.
The overview of the language pairs made up of the 23 European languages in the form of a large tabular matrix has given the project its name.