n-best reranking for the efficient integration of word sense disambiguation and statistical machine translation

Abstract

Although it has been always thought that Word Sense Disambiguation (WSD) can be useful for Machine Translation, only recently efforts have been made towards integrating both tasks to prove that this assumption is valid, particularly for Statistical Machine Translation (SMT). While different approaches have been proposed and results started to converge in a positive way, it is not clear yet how these applications should be integrated to allow the strengths of both to be exploited. This paper aims to contribute to the recent investigation on the usefulness of WSD for SMT by using n-best reranking to efficiently integrate WSD with SMT. This allows using rich contextual WSD features, which is otherwise not done in current SMT systems. Experiments with English-Portuguese translation in a syntactically motivated phrase-based SMT system and both symbolic and probabilistic WSD models showed significant improvements in BLEU scores. © 2008 Springer-Verlag Berlin Heidelberg.