Implementation of Word Alignment System for English- Myanmar Machine Translation

Abstract

Statistical word alignment models have been widely used for various Natural Language Processing (NLP) problem. In statistical machine translation, word alignment models are trained on bilingual corpora. To build an SMT system we require bitext and a word alignment of that bitext, as well as language models built from target language data. A word alignment for a parallel sentence pair represents the correspondence between words in a source language and their translations in a target language. This system will use the IBM model which is based on the EM algorithm. This system deals with the step of word alignment. In this paper, C# implementation of a word alignment algorithm is used to testing the source and target sentences. This system also uses a English-Myanmar dictionary to bootstrap the Expectation Maximization (EM) algorithm.