alvations / watercooler

Significant Machine Translation news/gossips...
2 stars 0 forks source link
machine-translation nlp

Overview

Google Translate

Ten years of Google Translate

  • 1. Google Translate helps people make connections.
  • 2. There are more than 500 million of you using Google Translate.
  • 3. Together we translate more than 100 billion words a day.

- Barak Turovsky (2016)

Microsoft Translate

"Microsoft reaches a historic milestone, using AI to match human performance in translating news from Chinese to English"

- Allison Linn (2018)

Facebook Translate

Facebook ditches Bing

  • "Facebook is now serving 2 billion text translations per day. ... when we turned it off for some people, they went nuts!”

- Josh Constine (2016)

eBay Translate

eBay’s Machine Translation Technology Breaks Down Borders

“Machine translation at eBay is key in promoting cross-border trade ... Our buyers can search the site in their native language, but see inventory from far away. ”

- Evgeny Matusov (2016)

Naver Translate

Amazon Translate

"To help our government customers and organizations in government-regulated industries, we are making Amazon Translate available in the AWS GovCloud (US) Region, Amazon’s isolated cloud region built for sensitive data and regulated workloads" - Woo Kim (2018)

Alibaba Translate

DeepL Translate

The Language Services Market

"People around the world prefer communicating in their native tongue. That fact will drive the language industry to US$40 billion in revenue for 2016."Common Sense Advisory (2016)

"Meeting the laws, needs, and expectations of inter-locale communication and commerce has fueled the language sector to 43 billion dollars in revenue." - Common Sense Advisory (2017)

"... this report analyzes the US$46.52 billion market from January 2017 through June 2018"Common Sense Advisory (2018)

The Volvo Incident

“In 2008, the Swedish car manufacturer Volvo was found partly to blame for a car accident which killed two school children. The expert engineer appointed by the court criticized the poor translation of a Third-Party Intermediary in the car manual on power-assisted brakes, the expert engineer stated, ‘... there is some cause for highlighting the fact ... that the technical product information EWP S 2631 (D 710) issued by the manufacturer and translated by the import was imprecise and poorly written‘. Volvo was fined 200,000 Euros for involuntary manslaughter and bodily injury . This called for a global concerns when handling terminology translations in technical manuals, especially in current translation workflows that incorporate machine translation with human post-editing.”

(c.f. Hoffmeister, 2013, Tan, 2016)

Google searches: https://www.google.com/search?q=volvo+2008+news+manslaughter&oq=volvo+2008+news+manslaughter

Managing expectation

"... it must be clearly emphasised that machine translation is neither a panacea nor an all-purpose tool for every translation task, which might thus be intended to eliminate jobs, but rather for general cost reduction. More importantly, it must be emphasised what machine translation cannot do, and why: for example, certain types of text are not suitable for machine translation or are (or could be) crucial for legal reasons."

- Jörg Porsiel (2017)

"those of us who have seen many paradigms come and go know that overgilding the lily does none of us any good"

- Andy Way (2018)

All your data belongs to me

"While most translators understand the importance of security, these concerns often take a back seat when a deadline is approaching. Translators should understand that all content uploaded to Google Translate or Bing Translator is used by those companies for future translation unit matches and language modeling. "

- Paul Walentynowicz (n.d.)

Governments want Translations

Countries

General

Crisis MT

Human level

"We should be able to do, with a modern computer, about 1 to 2 million words an hour. This would be quite an adequate speed to cope with the whole output of the Soviet Union. It's just a few hours computer time a week ... If our experiment goes well, [we'll be able to achieve the speed], in about 5 years or so.

"Yes, [it means the end of human translators], for translators for scientific and technical material but as regards poetry and novel, no I don't think we'll ever replace these translators"

- "The Thinking Machine" (1960s)

tl;dr Thanks to increasingly advanced neural network techniques and large amounts of data, translations by NMT systems can often sound fluent and human-like. But despite this seemingly impressive state of affairs, NMT is still far from reliable. - Sharon Zhou (2018)

Speech to Speech Translation

Translation and Business