On 09 Jun 2003, Jeffrey L. Taylor wrote: > Quoting stan <[EMAIL PROTECTED]>: > [snip] > > Looks like the maintainer of the Spambouncer rules is falling a bit > > behind the stae of the art of spammers, as I'm getting a fair number > > of spam emails into my primary mailbox each day. > > > > I'm thinking that prehaps spamassain will do better. Any opinions on this? > > > > More is getting thru SpamAssassin (SA) lately too. The spammers are > figuring out how to avoid SA's patterns. They are probably doing the > same for Spambouncer. Adding Bayesian filtering, either SA's or some > other package, helps, but requires maintenance of the spam and ham > (non-spam) samples. I have a setup with shared IMAP folders for false > positives and false negatives. Once an hour a cron job runs to feed > the corrections back into the Bayesian databases. > > SA is a real CPU and memory hog. I estimate a 266MHz Pentium as a > minimum. It absolutely choked a 133MHz 486 with 64MB RAM. > > HTH, > Jeffrey >
I'm getting pretty reasonable performance with spamprobe (a Bayesian filter). After a few weeks' use I'm getting only about 2 per cent of false negatives and almost no false positives. AC -- [EMAIL PROTECTED] || http://www.acampbell.org.uk using Linux GNU/Debian || for book reviews, electronic Windows-free zone || books and skeptical articles -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
