Prediction of Words in Statistical Machine Translation using a Multilayer Perceptron
Alexandre Patry, Philippe Langlais · 2009
We propose to estimate the probability that a target word appears in the translation of a given source sentence using a multilayer per-ceptron. At the expense of ignoring word order and repetition, our model does not as-sume word alignments and consider all source words jointly when evaluating the probability of a target word. We compared our model against IBM1 which does not consider word order either. Our model was comparable with IBM1 when pre-dicting the target words that should appear in the translation of a source sentence. When our model was extended to include alignment in-formation, it surpassed IBM1 on all the metrics we used. 1