Main Article Content
After availability of cheaper large memory and high performance processors, Statistical Machine Translation (SMT) methods have drawn attention of researchers NLP. Phrase-based SMT has shown better results than word-based SMT. To improve performance of machine translation system further, different systems have been developed which use phrase-based SMT as a baseline system. Domain adaptation is one of most popular example of such systems. In this paper also phrase-based SMT system is used as baseline to apply topic model for English-Hindi translation. This baseline system is also used for result comparison with topic model system. Both systems are trained using MERT. The analysis shows improvement in results obtained by using topic modeling system.