|Title||RNN-LDA AND LSTM CONTEXT MODELLING FOR MACHINE TRANSLATION|
Mani Bansal & Dr. D.K. Lobiyal
JNU, New Delhi, INDIA
Machine Translation is a primary issue while using machine learning that does not take into consideration dependency between words, sentences and phrases. By using Neural Network for translation resolves any problems. Therefore, this work presents two methods for Statistical translation of English-Hindi language pair. Firstly, we proposes Recurrent Neural Network (RNN) language model integrated with Latent Dirichlet Allocation (LDA) topic classification and after that Long Short-Term Memory (LSTM) method a special kind of Recurrent Neural Network to overcome long-term dependency. The long-span context of source and target side language motivates to apply these models on English-Hindi languages. It differentiates from the system that occupies the more memory and time consuming. The testing of sentences using BLEU score has given excellent results. By comparing the scores of these models gave equivalent output.