【摘 要】
：
The mission of Facebook is to make the world more open and connected and to give people the power to share.On this platform,every month one fifth of the peo
【机 构】
：
Tech Lead on Machine Translation Research Lead Research Scientist at Facebook
论文部分内容阅读
The mission of Facebook is to make the world more open and connected and to give people the power to share.On this platform,every month one fifth of the people in the world share their experiences using over 100 languages.However,existing language barriers limit the sharing and connection between people speaking different languages.In this talk,I will present how we are developing technologies and platforms at Facebook to break such barriers.Our technologies enable serving translations from authors,professional translators,and our community,as well as providing automatic translations,to help communication between people using different languages.I will describe various challenges we are facing,as well as our solutions for language identification,user language modeling,automatic translations(statistical and neural MT systems),human translation,and feedback,etc..I will conclude the talk with our vision of understanding Facebook contents in any language.
其他文献
In order to improve the efficiency of human translation,there is an increasing interest in applying machine translation(MT)to computer assisted translation(
HowNet is a Chinese-English Bilingual common-sense knowledge base,playing an important role in machine translation tasks.However,when fac-ing domain-specifi
科技术语翻译要求高度的准确性和专业性,通过建立术语语料的领域知识标签,并基于待翻译术语的领域对训练语料进行筛选,可训练出针对领域的翻译模型,能极大改善科技
显著性检验常用来判断系统之间的性能差异是否来源于系统的性能改善而不是随机误差。用于机器翻译系统的显著性检验通常以句子作为基本的抽样单位，忽略了抽样样本之间的独
机器翻译的发展有两个重要的维度，一是探索更为有效的数学工具对翻译的建模、计算等问题进行求解，二是使用丰富的先验知识来引导系统进行更加“合理”的翻译。我们发现二者
近几年来,一种基于词嵌入的线性变换方法在可比语料双语词典抽取任务上取得了显著的效果.这种方法假定双语词嵌入空间在翻译时满足线性关系,然而,在实际中,双语词嵌
机器翻译本质上是离不开语义的，它不仅需要解决源语言的语义表示问题，还要保证源语言的语义完整、正确地传递到目标语言上。最近几年我们可以看到机器翻译研究的兴趣正逐渐
基于数据驱动的机器翻译方法严重受限于双语训练数据的规模。最为直接的影响之一便是集外词翻译问题：如何处理训练语料中未出现过的词语。由于模型约束与计算复杂度的限制
基于Encoder-Decoder 的神经机器翻译模型将源语言句子编码到一个语义空间，并基于这个语义空间的源语言句子表示来产生目标语言句子。相比与传统的统计机器翻译，基于隐式语
In this talk,we will first give an overview of research on data selection for machine translation domain adaptation.Then,we will introduce a recently propos