Paper: Large-scale Expected BLEU Training of Phrase-based Reordering Models

ACL ID D14-1132
Title Large-scale Expected BLEU Training of Phrase-based Reordering Models
Venue Conference on Empirical Methods in Natural Language Processing
Session Main Conference
Year 2014
Authors

Recent work by Cherry (2013) has shown that directly optimizing phrase-based re- ordering models towards BLEU can lead to significant gains. Their approach is lim- ited to small training sets of a few thou- sand sentences and a similar number of sparse features. We show how the ex- pected BLEU objective allows us to train a simple linear discriminative reordering model with millions of sparse features on hundreds of thousands of sentences re- sulting in significant improvements. A comparison to likelihood training demon- strates that expected BLEU is vastly more effective. Our best results improve a hi- erarchical lexicalized reordering baseline by up to 2.0 BLEU in a single-reference setting on a French-English WMT 2012 setup.