Great, just wanted to be sure.


> -----Original Message-----
> From: Matt Kettler [mailto:[EMAIL PROTECTED] 
> Sent: Friday, January 23, 2004 9:38 AM
> To: Mark Squire; [EMAIL PROTECTED]
> Subject: Re: [SAtalk] Auto-learn SA after having trained it
> At 10:19 AM 1/23/2004, Mark Squire wrote:
> >Hi all,
> >I have been training SA manually for a couple of weeks now.  
> I estimate 
> >a good 2000 emails for both Spam and Ham have been learned by it. 
> >Coupla questions though . . . I want to put it into auto-learn mode 
> >because I have only trained it on a few of our employees emails, and 
> >not people from the whole company.  I think that SA needs to 
> "get out 
> >more" and learn from a broader range of emails (if that 
> makes sense).  
> >I wanted to be sure that it is okay to put it into auto-learn mode, 
> >even after I have been manually teaching it for a while.  
> What do you 
> >good folks think?
> Auto-learning is not mutually exclusive with manual training. 
> In fact, if 
> you are using auto learning, you SHOULD use manual training as well.
> Auto learning alone does NOT work, and will over time result 
> in a pretty 
> skewed bayes database. It needs some manual training as well.
> However, autolearning is quite useful, it's just not good 
> enough to be used 
> without ever training manually

The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
Spamassassin-talk mailing list

Reply via email to