Towards String-to-Tree Neural Machine Translation
This work addresses translation quality for users of NMT systems, but it is incremental as it builds on existing methods with a simple syntactic integration.
The authors tackled the problem of improving neural machine translation by incorporating target-language syntactic information, resulting in a higher BLEU score on the WMT16 German-English task compared to a baseline.
We present a simple method to incorporate syntactic information about the target language in a neural machine translation system by translating into linearized, lexicalized constituency trees. An experiment on the WMT16 German-English news translation task resulted in an improved BLEU score when compared to a syntax-agnostic NMT baseline trained on the same dataset. An analysis of the translations from the syntax-aware system shows that it performs more reordering during translation in comparison to the baseline. A small-scale human evaluation also showed an advantage to the syntax-aware system.