BriefGPT.xyz
Mar, 2022
尽可能少，尽可能多：使用对比调节检测过度翻译和欠翻译
As Little as Possible, as Much as Necessary: Detecting Over- and Undertranslations with Contrastive Conditioning
HTML
PDF
Jannis Vamvas, Rico Sennrich
TL;DR
本篇研究提出了一种用于检测神经机器翻译中内容省略和添加现象的方法，并使用对比学习比较翻译模型下完整序列的可能性和其部分的可能性，精确定位翻译中多余的词和源文本中未翻译的词，其准确性与需要自定义质量估计模型的监督方法相当。
Abstract
Omission and addition of content is a typical issue in
neural machine translation
. We propose a method for detecting such phenomena with off-the-shelf translation models. Using
contrastive conditioning
, we compar
→