Exploring Machine Translation on the Chinese-Vietnamese Language Pair (Extended Abstract)

Abstract

In the machine translation field, a common approach for more than two decades has been phrase-based statistical machine translation (PSMT). However, deep learning-based machine translation, also called neural machine translation (NMT), has emerged as a potential new approach to machine translation. Initial findings show that NMT yields better results than PSMT in some language pairs, while NMT results for some language pairs are less than or equal to PSMT results. In this paper, we initially studied and performed machine translation-based NMT for the Chinese-Vietnamese language pair. In addition, we performed machine translation-based PSMT on the same Chinese-Vietnamese bilingual corpus. The experimental results showed that NMT yielded better results (indicated by Bilingual Evaluation Understudy (BLEU) scores) than PSMT, although in some specific cases, NMT translation results were lower than those of PSMT.