When Google Translate was new in my language Finnish, their translation quality was atrocious.
I tried this sentence: "Hei äiti, puhun suomea". The expected translation would be "Hi mom, I speak Finnish".
Instead Google's result was: "July's mother, I speak English".
Obviously the engine had been trained on unvetted data sets where the word "English" occurred in translations in a position where the original had the word "Finnish", and no context was provided to avoid this kind of mistake.
The word "July" came about because "hei" is also used as an abbreviation for "heinäkuu" (July). It was sobering that a supposed world-class AI couldn't distinguish between these two usages. Machine learning needs a lot of old-fashioned handtuned human-made heuristics.
Sobering perhaps to learn what the state of the art is for "world class" AI indeed ;)
Cortana when??
Re:AWS Public Datasets
by solarkraft on Monday March 26, 2018 @06:02AM
(#16677790)
The German ML translation company of ex-googlers called DeepL (formerly Linguee) providing the currently best machine translation service apparently got a lot of their data from (I assume public) EU documents which had to be translated to all member languages.
They finally solved this by using meeting notes from UN Assembly. Which were transcribed by best of the translators? that access to meeting transcription was (unfair ??) advantage Google had over other tools. Was it wrong ? I don’t think so. Should have been those meeting notes be public: Yes