Achieving Open Vocabulary Neural Machine Translation with Hybrid Word-Character Models
This work addresses the limitation of restricted vocabularies in neural machine translation, particularly for languages with complex morphology like Czech, by providing a faster and more effective solution than purely character-based approaches.
The paper tackles the problem of open vocabulary neural machine translation by introducing a hybrid word-character model that translates at the word level while using character components for rare words, achieving a new state-of-the-art result of 20.7 BLEU score on the WMT'15 English to Czech task with improvements of +2.1-11.4 BLEU points over existing models.
Nearly all previous work on neural machine translation (NMT) has used quite restricted vocabularies, perhaps with a subsequent method to patch in unknown words. This paper presents a novel word-character solution to achieving open vocabulary NMT. We build hybrid systems that translate mostly at the word level and consult the character components for rare words. Our character-level recurrent neural networks compute source word representations and recover unknown target words when needed. The twofold advantage of such a hybrid approach is that it is much faster and easier to train than character-based ones; at the same time, it never produces unknown words as in the case of word-based models. On the WMT'15 English to Czech translation task, this hybrid approach offers an addition boost of +2.1-11.4 BLEU points over models that already handle unknown words. Our best system achieves a new state-of-the-art result with 20.7 BLEU score. We demonstrate that our character models can successfully learn to not only generate well-formed words for Czech, a highly-inflected language with a very complex vocabulary, but also build correct representations for English source words.