[Corpora-List] Google's translations

[Tad Piotr]

The quality of translation, between Google and Bing, must be heavily
dependent on the language pair. The bit about cost of oil, when
translated into Polish, produces nonsense in Bing, and the result is
understandable in Google.
Tadeusz

On 15 March 2010 12:41, John F. Sowa <sowa at bestweb.net> wrote:
>> Clearly the massive amount of English training data has certain
>> extremely unfortunate consequences for the choice of lexical items.
>> It is amusing, but could be potentially very confusing, when one
>> currency is translated to another (but not recalculated!).,,
>>
>> Olje koster kr 800 fatet. Gassen koster NOK 400 mens vindkraft
>> koster seksti kroner.
>>
>> Oil costs U.S. $ 800 per barrel. The gas costs NOK 400, while
>> wind power costs sixty kroner.
>
> That is an example where some rule-based processing should be
> added as a supplement to statistical methods.  It could recognize
> pairs of (unit, value) and either translate both or leave both
> unchanged.
>
> Wolfram's Alpha system (which is used in Microsoft's Bing) can
> handle such examples.  I typed "What is 400 NOK in dollars?"
> to Bing.com and got the answer
>
>   400 NOK = 68.38800 USD
>
> For "What is sixty kroner in dollars?", Wolfram didn't do anything,
> but the first two hits in Bing were web sites for converting
> Danish kroner and Norwegian kroner.
>
> For "What is kr 800 in dollars?" the first hit was a web site
> with the title "What is 800 euros in dollars? - Yahoo! Answers"
>
> John Sowa
>
>
> _______________________________________________
> Corpora mailing list
> Corpora at uib.no
> http://mailman.uib.no/listinfo/corpora
>



-- 
Tad Piotr

_______________________________________________
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora