AUGMENTING STATISTICAL MACHINE TRANSLATION WITH LINGUISTIC KNOWLEDGE

摘要

Methods and systems for improving phrase-based statistical machine translation, particulary from a non-morphologically rich language (e.g. English)to a morphologically rich language (e.g. Arabic). The first part of the specification aims at improving morphology by adding features to a phrase table and thus modifying the main pipeline of the statistical machine translation. The second part of the specification introduces a post-processing framework for fixing inflectional agreement in machine translation output. The specification focuses particularly on morphological agreement be¬ tween syntactically dependent words.