In this paper, we enhance the attention-based neural machine translation (NMT) by adding explicit coverage embedding models to alleviate issues of repeating and dropping translations in NMT. For each source word, our model starts with a full coverage embedding vector to track the coverage status, and then keeps updating it with neural networks as the translation goes. Experiments on the large-scale Chinese-to-English task show that our enhanced model improves the translation quality significantly on various test sets over the strong large vocabulary NMT system.

In this paper, we describe a new reranking strategy named word lattice reranking, for the task of joint Chinese word segmentation and part-of-speech (POS) tagging. As a derivation of the forest reranking for parsing (Huang, 2008), this strategy reranks on the pruned word lattice, which potentially contains much more candidates while using less storage, compared with the traditional n-best list reranking. With a perceptron classifier trained with local features as the baseline, word lattice reranking performs reranking with non-local features that can't be easily incorporated into the perceptron baseline. Experimental results show that, this strategy achieves improvement on both segmentation and POS tagging, above the perceptron baseline and the n-best list reranking.

In this paper, we improve the attention or alignment accuracy of neural machine translation by utilizing the alignments of training sentence pairs. We simply compute the distance between the machine attentions and the "true" alignments, and minimize this cost in the training procedure. Our experiments on large-scale Chinese-to-English task show that our model improves both translation and alignment qualities significantly over the large-vocabulary neural machine translation system, and even beats a state-of-the-art traditional syntax-based system.

Translation rule extraction is a fundamental problem in machine translation, especially for linguistically syntax-based systems that need parse trees from either or both sides of the bitext. The current dominant practice only uses 1-best trees, which adversely affects the rule set quality due to parsing errors. So we propose a novel approach which extracts rules from a packed forest that compactly encodes exponentially many parses. Experiments show that this method improves translation quality by over 1 BLEU point on a state-of-the-art tree-to-string system, and is 0.5 points better than (and twice as fast as) extracting on 30best parses. When combined with our previous work on forest-based decoding, it achieves a 2.5 BLEU points improvement over the baseline, and even outperforms the hierarchical system of Hiero by 0.7 points.