Entrenamiento y comparativa de motores de TAE especializados en la localización de aplicaciones móviles

The quality of statistical machine translation (SMT) is closely related to the characteristics of the corpus used, such as its volume and its homogeneity. This article describes the training process of three statistical machine translation engines with different collections of texts combi- ned, all specialized in the translation of mobile apps. The three engines will be compared in order to determine which composition of corpus is the best for training engines specialized in app localisation, and to check how the trained engines deal with some problematic aspects related to this type of translation.