Pre-reordering for machine translation using transition-based walks on dependency parse trees

We propose a pre-reordering scheme to improve the quality of machine translation by permuting the words of a source sentence to a target-like order. This is accomplished as a transition-based system that walks on the dependency parse tree of the sentence and emits words in target-like order, driven by a classifier trained on a parallel corpus. Our system is capable of generating arbitrary permutations up to flexible constraints determined by the choice of the classifier algorithm and input features.