Monday, January 31, 2011

[bhqgoamw] CAPTCHA based on public spam

Create a CAPTCHA system similar to reCAPTCHA, but instead of text which the OCR couldn't read, we provide text that the spam classification algorithm couldn't decide on.

Some public sources include Wikipedia (automatically detecting spam and vandalism) and forum and comment spam.  Obviously these human solved examples can be used to further train the spam classifier.

An incestuous use case might be to require (or maybe optional) solving such a CAPTCHA before submitting a Wikipedia edit.

1 comment :

Anonymous said...

wow..........grt to know the captchas were created based on public scams.................


death by captcha