Comparison of Statistical Approaches for Tamil to English Translation

This work proposes a Machine Translation system from Tamil to English using a Statistical Approach. Statistical machine translation (SMT) is a machine translation paradigm where translations are generated on the basis of statistical models whose parameters are derived from the analysis of bilingual text corpora. It is the most widely used machine translation paradigm for the tradeoff between efficiency and implementation feasibility and due to its partial language independency. In syntax based approach, a phrase table is created which identifies the most probabilistically likely English translation of each Tamil phrase in the input sentence. In hierarchical phrase based approach, a rule table is used to reduce the input Tamil sentence into the output English sentence. We evaluated the two approaches based on different parameters like corpus size, gram size of language model and achieved a BLEU score of 0.26.