A Neural Network based Approach for English to Hindi Machine Translation

Abstract

In this paper we are discussing the working of our English to Hindi Machine Translation system. Our system is able to translate English language’s simple sentences into Hindi. This system has been implemented using feed-forward backpropagation artificial neural network. ANN model does the selection of translation rules for grammar structure and Hindi words/tokens (such as verb, noun/pronoun etc.). Neural network is used as the knowledge base and for mapping process from bilingual dictionary and linguistic rules. Bilingual dictionary is implemented using neural network, stores the meaning and linguistic features attached to the word of English and Hindi. The transformation of one natural language grammar to other natural language is the core of the machine translation specifically when the languages have different grammatical class such English and Hindi. Grammatical Structure analysis is done with the help of Stanford Tagger and Stanford Parser. The developed module is able to translate simple sentence of English language. The evaluation score achieved by the system for around 500 test sentences is: n-gram blue score 0.604; METEOR score achieved is 0.830 and F-score of 0.816. General Terms Natural language processing, machine translation