Improving sampling-based alignment by investigating the distribution of n-grams in phrase translation tables

Juan Luo (1) , Adrien Lardilleux (2) , Yves Lepage (1)

1 IPS - Graduate School of Information, Production and Systems
2 LIMSI - Laboratoire d'Informatique pour la Mécanique et les Sciences de l'Ingénieur

This paper describes an approach to improve the performance of sampling-based multilingual alignment on translation tasks by investigating the distribution of n-grams in the translation tables. This approach consists in enforcing the alignment of n-grams. The quality of phrase translation tables output by this approach and that of MGIZA++ is compared in statistical machine translation tasks. Significant improvements for this approach are reported. In addition, merging translation tables is shown to outperform state-of-the-art techniques.

Mots clés: alignment; phrase translation table; statistical machine translation

Domaines: Traitement du texte et du document; Informatique et langage [cs.CL]

Juan Luo, Adrien Lardilleux, Yves Lepage. Improving sampling-based alignment by investigating the distribution of n-grams in phrase translation tables. 25th Pacific Asia Conference on Language, Information and Computation (PACLIC 25), Dec 2011, Singapore. pp.150-159. ⟨hal-00747369⟩