Show HN: My Leeds Hack Day project, a language detection API

Post by zorander » Sun, Nov 07, 2010, 8:53 PM UTC

Post by zbanks » Sun, Nov 07, 2010, 9:02 PM UTC

lgeek wrote: post #3

zorander wrote: post #2

Uhhh..what? http://imgur.com/LH2Hi.png

Sorry, only supports English, German and French at the moment

Post by catch23 » Sun, Nov 07, 2010, 10:02 PM UTC

lgeek wrote: post #5

zbanks wrote: post #4

Earlier quoted context omitted.

"Je t'aime" is wrongly classified as English. (Although understandable, since it's short) How does this work? You claim it has "zero language knowledge," so how does it classify? Did it start out with nothing, and then train it on a corpus? EDIT: nevermind, found your slides ( http://polyglossy.com/presentation.pdf [pdf]) How big of a corpus did you use?

I've used around 40KB of text per language for training.

Post by zbanks » Sun, Nov 07, 2010, 11:50 PM UTC

lgeek wrote: post #7

catch23 wrote: post #6

Earlier quoted context omitted.

wouldn't something like 2-gram bayes be more accurate & faster?

I don't know. I haven't really done NLP before. My guess is that it would be slower, but not by a significant amount. It might be more accurate, so I could give it a shot.

Post by lgeek » Mon, Nov 08, 2010, 1:29 PM UTC

zbanks wrote: post #8

lgeek wrote: post #7

Earlier quoted context omitted.

I don't know. I haven't really done NLP before. My guess is that it would be slower, but not by a significant amount. It might be more accurate, so I could give it a shot.

Any chance of posting the source?