A Novel Real-Time Machine Translation System based on Cloud LM Algorithm

This research work proposes a novel real-time Japanese machine translation system based on Cloud LM algorithm. The proposed model focuses on 2 aspects, namely: 1. The novel Japanese machine translation algorithm is designed and modeled. 2. The efficient cloud computing framework is designed with the specialized analysis on the data transmission is studied to achieve the real-time translation system. Fort the algorithm design part, the 3-step approach is proposed: (1) the pre-training model for the Transformer model is designed. (2) improved byte pair encoding model considering the Japanese semantic mining is studied. (3) The neural machine translation model based on the hierarchical analysis is considered. Then, the parasitic transmission channel model is designed for the real-time cloud system. The proposed system is validated through the test on the accuracy and cloud response time.