Achieving Human Parity on Automatic Chinese to English News Translation
Hany Hassan Note: Corresponding author: Affiliation: Microsoft AI、Research Anthony Aue Affiliation: Microsoft AI、Research Chang Chen Affiliation: Microsoft AI、Research Vishal Chowdhary Affiliation: Microsoft AI、Research Jonathan Clark Affiliation: Microsoft AI、Research Christian Federmann Affiliation: Microsoft AI、Research Xuedong Huang Affiliation: Microsoft AI、Research Marcin Junczys-Dowmunt Affiliation: Microsoft AI、Research William Lewis Affiliation: Microsoft AI、Research Mu Li Affiliation: Microsoft AI、Research Shujie Liu Affiliation: Microsoft AI、Research Tie-Yan Liu Affiliation: Microsoft AI、Research Renqian Luo Affiliation: Microsoft AI、Research Arul Menezes Affiliation: Microsoft AI、Research Tao Qin Affiliation: Microsoft AI、Research Frank Seide Affiliation: Microsoft AI、Research Xu Tan Affiliation: Microsoft AI、Research Fei Tian Affiliation: Microsoft AI、Research Lijun Wu Affiliation: Microsoft AI、Research Shuangzhi Wu Affiliation: Microsoft AI、Research Yingce Xia Affiliation: Microsoft AI、Research Dongdong Zhang Affiliation: Microsoft AI、Research Zhirui Zhang Affiliation: Microsoft AI、Research Ming Zhou Affiliation: Microsoft AI、Research
Abstract
Machine translation has made rapid advances in recent years. Millions of people are using it today in online translation systems and mobile applications in order to communicate across language barriers. The question naturally arises whether such systems can approach or achieve parity with human translations. In this paper, we first address the problem of how to define and accurately measure human parity in translation. We then describe Microsoft’s machine translation system and measure the quality of its translations on the widely used WMT 2017 news translation task from Chinese to English. We find that our latest neural machine translation system has reached a new state-of-the-art, and that the translation quality is at human parity when compared to professional human translations. We also find that it significantly exceeds the quality of crowd-sourced non-professional translations.
原文 arXiv:1803.05567;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1803.05567v2