just how much detail do they use?
Japanese translator:
"Picture is 1024X750, 98% black with white oval smudge near center."
English translator compares picture of galaxy with colorful flower picture:
"That would be the picture on the left."
Computers are getting pretty good at translating the world's languages. However, as they say, onwards and upwards. Eggheads are now trying to teach machines to do the job in a more human-like way. “While most machine translation systems to date are trained on large parallel corpora, humans learn language in a different way: by …
Lets gloss over the more disturbing truth. Don't look deeper... Reminder of the monster lurking below that Facebook would prefer we forget about:
https://www.nytimes.com/2016/11/20/opinion/the-secret-agenda-of-a-facebook-quiz.html?_r=1
https://www.bloomberg.com/view/articles/2017-10-27/that-facebook-quiz-might-not-be-so-innocent
"Computers are getting pretty good at translating the world's languages."
No, they really aren't. I regualrly use Google translate to read a German language website and have to use domain knowledge and my terrible school German from almost 30 years ago to re-translate what Google provides into the correct English. I'm not talking about cleaning up the grammar here, I'm talking about correcting sentences which are either total garbage or say the opposite of what was written.
"it’s nowhere near as good as Google Translate nor Facebook’s translation system"
Facebook translate is so good it translates "good morning" into "attack them": https://www.theguardian.com/technology/2017/oct/24/facebook-palestine-israel-translates-good-morning-attack-them-arrest
@Irongut
I was about to post exactly the same thing. Roughly a third of the time when I'm using some auto translation service (Google is worse than others admittedly) from Dutch to English it adds or removes a negative in the sentence. For train tickets it almost always translates "may only be used after 0900" to "may not be used after 0900". That's from simple sentences in standard Dutch.
It's something that should be easily available to check with many other examples online, the terms and conditions are often available in English too, so I can't understand how it's still failing to get simple translations exactly reversed.
It's also pretty iffy for translating Dutch legalese, with again many switching of negatives, but I accept that there are not always going to be decent examples for my particular case.
Relying on auto translate without any context is asking for trouble. I've also not found many ML/AI/programs that can take a statement and context and provide any sort of subtext. So how translating meaning is going to work I'm at a loss.