3 ms·
To help with all the many postings expressing surprise at a misidentification: How long was your sample? According to the link posted by fixz, http://alias-i.co
by jerf 17y ago
To help with all the many postings expressing surprise at a misidentification: How long was your sample? According to the link posted by fixz, http://alias-i.com/lingpipe/demos/tutorial/langid/read-me.html http://alias-i.com/lingpipe/demos/tutorial/langid/read-me.ht... , this is using letter-frequency statistics. It's very easy to fool such a system by feeding it a very short sample. For a similar problem that can arise with trying to auto-detect encodings, see: http://blogs.msdn.com/oldnewthing/archive/2004/03/24/95235.aspx http://blogs.msdn.com/oldnewthing/archive/2004/03/24/95235.a...
Try feeding it a bit more than a single word or sentence.
- andreaolivato 17y agoYeah you got that right. The more you type the more google apis should be accurate.