Simultaneous Neural Machine Translation with Constituent Label Prediction
This work addresses the problem of reducing latency in real-time translation for users of language pairs with divergent word orders, though it is incremental as it builds on pre-reordering concepts.
The paper tackled the challenge of deciding when to start translating in simultaneous neural machine translation for language pairs with different word orders, such as English-to-Japanese, by proposing simple decision rules based on predicted constituent labels, resulting in improved quality-latency trade-off compared to baselines.
Simultaneous translation is a task in which translation begins before the speaker has finished speaking, so it is important to decide when to start the translation process. However, deciding whether to read more input words or start to translate is difficult for language pairs with different word orders such as English and Japanese. Motivated by the concept of pre-reordering, we propose a couple of simple decision rules using the label of the next constituent predicted by incremental constituent label prediction. In experiments on English-to-Japanese simultaneous translation, the proposed method outperformed baselines in the quality-latency trade-off.