Role Of Machine Learning In Machine Translation

Abstract

To enhance the quality of machine translation (MT), the most significant way is to train the system. Machine Learning (ML) provides an opportunity to train the system by using a training data set. Further, assessment is done using a test data set. This way training and assessment can be carried out in a given system. The training further helps the system make appropriate predictions. To provide training and testing data ML has techniques like supervised, unsupervised, and knowledge-based. MT has various challenges. Including ambiguity. In this paper, we have done an analysis of various ML techniques for Word Sense Disambiguation (WSD). The accuracy of all the analyzed algorithms ranges from sixty-eight to eight-four percent. We have also introduced a hybrid model named AmbiF to resolve WSD. This model has exhibited a higher accuracy percentage in comparison to the other analyzed techniques. The percentage of accuracy reported is eighty-five percent under the F-score value.