A-priori identification of the MT systems that will produce the best translations
The goal of this project is to determine in advance, without translating, the subset of MT systems, among a larger set of MT systems, that is most likely to produce the best translations by using only information extracted from the source language sentence to translate, i.e. without any information about the inner workings of the MT systems. A system able to select in advance the systems to use to translate a sentence will allow multi-engine MT systems to save computing resources and focus on the combination of the best translations. At the same time, such a tool will allow to increase the number of MT systems in which current multi- engine MT systems are based.
