Thursday, 8. January 2004
kSpamLoaded kSpam onto a test server here this morning and trained its Bayesian filter with over 3,000 items of spam and a somewhat smaller but representative number of items of good email.
So far so good. I threw a copy of one of the known spam messages at the server and got:
08/01/2004 12:05:41 kSpam Bayesian: Message from '"nova-light.com" <nova-light@worldnet.att.net>' copied & denied.
Very encouraging. Then I threw a freshly minted mortgage spam that hit a spamtrap here only moments ago at the same server and it strolled right past the filters and hit my mailbox. Guess I need to spend some more time training the Bayesian thing. Looking forward to Francois Lafleur's promised next lesson on kSpam.
Update 22 March 2005: That promised next lesson never materialised and in fact the post linked to on François Lafleur's site was his last ever post. That site may still exist but the ACL has been changed to deny anonymous access and so you can no longer read any of Mr Lafleur's commentary on KSpam, which is a pity.
Category: Software
Technorati: Software
Comments :
Unable to post a comment? Please read this for a possible explanation...
- 
