Comparison of machine translations (MT) technology; statistical (SMT) vs. neural (NMT)

The objectives of this study are to compare two types of machine translations (MT) namely statistical machine translations (SMT) and neural machine translations (NMT), and to analyse the accuracy of the two types. SMT uses statistical model of bilingual text corpora while NMT uses artificial neural network to produce output in target languages. Variants in MT were found in this study, supporting previous studies on variants in translations. Comparing machine translations with human translations, the former did no better than the latter. Human translations is, despite being costly and timely, still more accurate than any MT available today. Having said that, it is not impossible for people to rely completely on machine translations in the near future as MT is being actively investigated studied for its betterment. Future studies might want to look at the implications on human translators towards their professions since the advancement of MT is astounding.