QCRI-MES Submission at WMT13: Using Transliteration Mining to Improve Statistical Machine Translation
Hassan Sajjad, Svetlana Smekalova, Nadir Durrani, Alexander Fraser, Helmut Schmid · 2013
This paper describes QCRI-MES’s submission on the English-Russian dataset to the Eighth Workshop on Statistical Machine Translation. We generate improved word alignment of the training data by incorporating an unsupervised transliteration mining module to GIZA++ and build a phrase-based machine translation system. For tuning, we use a variation of PRO which provides better weights by optimizing BLEU+1 at corpus-level. We transliterate out-of-vocabulary words in a postprocessing step by using a transliteration system built on the transliteration pairs extracted using an unsupervised transliteration mining system. For the Russian to English translation direction, we apply linguistically motivated pre-processing on the Russian side of the data.