Let's face it: data mining isn't perfect. Sure, it seems easy: plug in some data, choose a fancy algorithm and, hey presto, predictions! Things can go wrong, however. Sometimes, disastrously wrong. Things like power law failures, BLAST accidents, decision tree forest fires, and even numeric overflow:
See Mary McGlohon's report linked below for all the potentially catastrophic pitfalls and some (fashion-forward!) recommendations for addressing them.
Mary McGlohon: Data Mining Disasters: a report (via Gregory Piatetsky)



Data mining is a most common technique for avoiding captcha text system. It is basically used for type written text.
Posted by: Ink | October 13, 2009 at 01:54