Focusing on BLEU Can Bias Machine Translation Output
Though beam search can boost BLEU scores, it can also lead to high rates of misgendered pronouns, even when translating between two gendered languages.
Though beam search can boost BLEU scores, it can also lead to high rates of misgendered pronouns, even when translating between two gendered languages.
Group of leading researchers finds that the way machine translation quality is evaluated by humans is broken. Group issues five recommendations on how to fix it.
From technology, markets, and M&A to investment and the competitive landscape, the language industry leaders we interviewed in 2019 commented on a wide range of issues. Here are our top picks.
Less than one year after Google open-sourced its much-discussed language model BERT, experts weigh in on its potential uses in neural machine translation.
New research by Google AI explores multilingual NMT on an unprecedented scale. Good for low-resource languages, less so for the rest, results show.
Real-time terminology assistance for simultaneous interpreters may be well on its way — but will it be truly useful in typical high-stakes environments?
Top US university Carnegie Mellon, known for its natural language processing and neural machine translation research, has launched a new Master’s program in Global Communication and Applied Translation, designed to train translators of the future.
US researchers outline how technology could be used to support the work of simultaneous interpreters. They propose analyzing the linguist’s performance on the fly via real-time quality feedback loop and offer help only when the interpreter is struggling.
Research into statistical machine translation is fast falling out of favor among the research community as the number of papers published on neural machine translation skyrockets in 2016.
Join over 15,400 subscribers and get the latest language industry intelligence every Friday
Your information will never be shared with third parties. No Spam.
This will close in 0 seconds