Machine Translation for all European Language Pairs
The EuroMatrix project aims to promote machine translation research for all pairs of official languages of the European Union. As part of this effort, this web site presents (a) reference translation performance for standard test sets, submitted by research groups and commercial machine translation providers, and (b) information about available resources for building machine translation systems.
This allows you to answer questions such as:
- How good is machine translation for any European language pair?
- Who is developing machine translation technology in Europe?
- What language resources are available?
The EuroMatrix project is funded by the EU Information Society Technology programme.
The EuroMatrix project organizes an annual evaluation campaign to bring together machine translation researchers. Participants in the campaign translate test sets with their systems, which then are evaluated using manual and automatic metrics. The systems are presented in a workshop which is co-located with a conference of the Association for Computational Linguistics.
This web site extends the evaluation campaign online: any system developer may translate a test set and upload their translations. The submitted set is automatically scored and displayed alongside the other.
See for more information:
Building machine translation systems requires many language resources such as parallel corpora (translated texts), linguistic tools such as morphological analyzers, part-of-speech taggers, parsers, etc. This web sites lists such available resources, along with existing machine translation systems.
Currently, the listing of resources is based on the Compendium of Translation Software, compiled by John Hutchins. Soon you will be able to add any additional resource you are aware of.