by Wolfram Saringer  (2010-01-27)
last change: 2010-01-27

Nice approach to filtering spam -- actually something I do manually right now. This filter collects multiple messages and tries to find the underlying template which is 'randomized' by words inserted from a limited-size dictionary. It reaches less than 1% false positives after gathering 100 spam-mails and gets even better with more... Might put some additional work on the spammers to evade this.

all articles represent the sole opinion of their respective author. all content comes without any warranty for correctnes, despite due diligence.