Multiword Expressions in the Context of Statistical Machine Translation

Incorporating semantic information in the Statistical Machine Translation (SMT) framework is starting to gain some popularity in both the semantics and translation communities. In this paper, we present encouraging results obtained from experiments conducted on English to Arabic SMT system using static, dynamic, and hybrid integration of fine-grained Multiword Expression (MWE). We achieve an improvement up to 0.82 absolute BLEU score by integrating MWEs over a vanilla SMT system. We empirically show that different MWE types require different integration methods in the SMT framework.