A Reordering Model for Phrase-Based Machine Translation
説明
This paper presents a new method for reordering in phrase based statistical machine translation (PBSMT). Our method is based on previous chunk-level reordering methods for PBSMT. Our method is a global reordering. First, we parse the source language sentence to a chunk tree, according to the method developed by [1]. Second, we apply a series of transformation rules, which are learnt automatically from the parallel corpus to the chunk tree over chunk level. Finally, we solve phenomena for the overlapping of phrases and chunks, and integrate a global reordering model directly in a decoder as a graph of phrases. The experimental results with English-Vietnamese and English-French pairs show that our method outperforms the baseline PBSMT in both accuracy and speed.
