Empirical survey of machine translation tools

Machine Translation (MT) has progressively evolved since 1940's. It is a topic of active research now a days as the results found so far from machine translation tools are very unrealistic as compared to the human translation. Many different new approaches and techniques have evolved along with the new advent in machine translation. There are different paradigm of machine translation including Statistics Based Machine Translation (SBMT), Rule Based Machine Translation(RBMT), Hybrid machine translation(HMT). Besides these, Neural Network Based Systems have been developed for machine translation [1]. We have not yet imparted the human kind of translation capabilities to the machine. Various online MT tools when tested on various input paragraphs from literature, though performed remarkably good but could hardly translate the sentences comparable to us, the humans. This paper provides a comparative study based on the translation of paragraphs by various online machine translation tools. The tools tested for this research involves rule based systems(Angla Bharti and Anubaad), and statistical systems (Bing, Google translator, IM translate that is supported by Microsoft Translator, Google Translate, Babylon Translator and other MT engines). The results shows that though statistical MT systems outperform the rule based machine translation, but as of yet human mankind is far from achieving its dream of creating a “perfect” automatic translation tool.