Recurrent neural network language model for English-Indonesian Machine Translation: Experimental study

At recent time, the statistical based language model and neural based language model are still dominating the researches in the field of machine translation. The statistical based machine translation today is the fastest one but it has a weakness in term of accuracy. In contrast, the neural based network has higher accuracy but has a very slow computation process. In this research, a comparison between neural based network that adopts Recurrent Neural Network (RNN) and statistical based network with n-gram model for two-way English-Indonesian Machine Translation (MT) is conducted. The perplexity value evaluation of both models show that the use of RNN obtains a more excellent result. Meanwhile, Bilingual Evaluation Understudy (BLEU) and Rank-based Intuitive Bilingual Evaluation Score (RIBES) values increase by 1.1 and 1.6 higher than the results obtained using statistical based.