http://www.ritlabs.com New posts in bayesit false positives of forum at www.ritlabs.com [www.ritlabs.com] en http://backend.userland.com/rss2 Wed, 06 Dec 2006 12:06:17 +0200 bayesit false positives in forum The Bat! - Configuring the E-mail Client.
let bayesit learn MANY "good" mails! ideally when starting with the filter more ham than spam words should be in your database (at least it should be about the same amount - lateron the spam dictionary may start to outnumber the ham dictionary as many 'senseless' words from headers or random spam will be added... but for meaningful words the stress should still be on the ham side)

(the sender is only a bit of the whole message, but bayes filter check all words in it... wanted content is the more important thing to learn!)
06 December 2006 12:06:17, nlkes.]]> http://www.ritlabs.com/en/forums/forum4/topic3890/message15004/ http://www.ritlabs.com/en/forums/forum4/topic3890/message15004/ Wed, 06 Dec 2006 12:06:17 +0200 The Bat! - Configuring the E-mail Client bayesit false positives in forum The Bat! - Configuring the E-mail Client.
The bayesit plug is doing a great job with learning and flagging spam, but it is also repeatedly giving me many false positives. I repeatedly mark emails from the same senders as Not Junk but it doesn't seem to help. I hate to keep having to add whitelist rules as that sort of defeats the purpose of teaching it what is Not Junk, no?

Ideas?
05 December 2006 07:00:35, z dub.]]> http://www.ritlabs.com/en/forums/forum4/topic3890/message14982/ http://www.ritlabs.com/en/forums/forum4/topic3890/message14982/ Tue, 05 Dec 2006 07:00:35 +0200 The Bat! - Configuring the E-mail Client