At Sun Aug 10 12:49:31 2003, Gerry Doris wrote:
> 
> I don't understand why you need to run this against the normal non-spam 
> mail.  Shouldn't all the ham have already passed through spamassassin and 
> be "learned".  

No.  When it comes to auto-learning, there are three classifications
for messages - "learn as spam", "learn as ham", "don't learn".
There's a significant safety margin between auto-learning spam and
auto-learning ham, and within this margin the message is not
auto-learnt. 

The logic behind this is to ensure that high-scoring non-spam is not
learnt as spam, and that low-scoring spam is not learnt as non-spam.
Auto-learning is very conservative, because mistakes can have an
effect on the effectiveness of Bayesian filtering.

> I manually move any spam that slipped through to my current mail to a spam
> mailbox and then later do an sa-learn as spam on it.  I've never bothered
> to learn the ham or the spam that spamassassin has already passed
> correctly.

That's one approach, but it's not using the Bayesian capabilities to
the full.  Messages that SA is tagging as spam don't necessarily get
autolearnt. 

Martin
-- 
Martin Radford              |   "Only wimps use tape backup: _real_ 
[EMAIL PROTECTED] | men just upload their important stuff  -o)
Registered Linux user #9257 |  on ftp and let the rest of the world  /\\
- see http://counter.li.org |       mirror it ;)"  - Linus Torvalds _\_V


-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01
_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk

Reply via email to