Improving the Quality of Machine Translation through Proper Transliteration of Name Entities

Abstract

Machine Translation is the study of system that translates the given input text in source language to the output text in the target language. The source language and the target language are Natural Languages. Machine Translation is a very difficult problem; especially name entity translation has always been a challenge for the machine translators because of different spelling variations in translation of name entities. In this paper we focus on improving the quality of our machine translated output. For this, at first we recognize the name entities and then transliterate them. We have calculated the phoneme based N-Gram Probabilities for all the name entities. Using these probabilities we are transliterating our name entities from English to Punjabi through syllabification in which we divide the word in to syllables.