(29) Improvement of Translation Quality by Application of Translation Rules Based on Generalization of Bilingual Sentences to Phrase-Based SMT
Ryo Terashima, Hiroshi Echizen-ya, Kenji Araki
Proceedings of the Pacific Association For Computational Linguistics 2011,2011-07
A phrase-based statistical machine translation (SMT) system
can generate translation results for variousinput sentences. However,
the phrase-based SMT system may generate unnatural translation results,
because it cannot deal with structures of sentences sufficiently. On
the other hand, in translation methods using translation rules based on
generalization of bilingual sentences, a system acquires translation
rules that possess structures of sentences. In this paper, we propose a
method that combines phrase-based SMT with translation rules based on
generalization of bilingual sentences. We are aiming at the improvement
of translation quality by using acquired translation rules without any
analytical tools in the phrase-based SMT. Evaluation experiments
indicated that the translation quality improved by combining phrasebased SMT with
translation rules based on generalization of bilingual sentences.