Exploration of Chinese-Uyghur neural machine translation

Nowadays two people who speak different languages are able to communicate with real-time translation software. This is benefited from machine translation technology. In China, there are multiple languages with great diversity. Uyghur and Chinese are the official languages of Xinjiang Uyghur Autonomous Region, China, which makes it urgent to improve the quality of Chinese-Uyghur (Uyghur-Chinese) machine translation. Recently, Neural machine translation (NMT) has reached promising results for most language pairs. Therefore, in this work, we first briefly analyze the difficulties of Uyghur machine translation. And then study the performance of Chinese-Uyghur machine translation with a statistical framework (PBMT) and two neural network frameworks (NMT and M-NMT), respectively. As a result, we not only have a better understanding of Chinese-Uyghur machine translation but also get our baseline system.