Syntax based machine translation using blended methodology

Abstract

The basic concept of machine translation is to translate one human language to another human language. Many translation systems have been built using different approaches which have different accuracy levels. In this paper, a work is being proposed for Syntax Based Machine Translation System from English to Hindi language. The Syntax Based Machine Translation System has the goal of incorporating an explicit representation of syntax into the statistical systems, to get the best out of the two worlds. The approach used in this translation system is a Blended approach i.e. statistical approach along with the Rule-based approach with the intention to get more accurate translation. Incorporating the idea of Synchronous Tree Substitution Grammar (STSG), it has been tried to parallel generate English and Hindi rules dynamically, which leads to final translated Hindi text. A parallel aligned text corpora of English and Hindi is being used as Database of the system, which is limited and hence limits the translation. Further, the system undergoes through various stages of Preprocessing, Source language analysis and then source to target language analysis to get the translation. The user has provision to select the best suitable translation according to him. This paper proposes the working strategy of how translation is being carried out from English to Hindi and the accuracy level that it achieved.