High Positive SA AWL rule ?

[Anthony Peacock]

Hi,

I am running with AWL and have had no major problems for a while now.

I think the trick with AWL is to only switch it on once you have a 
stable setup and you are generally happy with your hit rate. If you 
use Bayes for instance, don't switch on AWL until your Bayes database 
is properly trained.  This seems to be true with the Bayes autolearn 
feature as well.

I am using SpamAssassin 3.1.0 with sitewide Bayes and AWL switched 
on.

> On Thu, September 29, 2005 06:22, Drew Marshall wrote:
> > On 28 Sep 2005, at 21:08, Matt Kettler wrote:
> >
> >> Mailscanner wrote:
> >>
> >>> Has anyone noticed SA run AWL giving high positive scores like;
> >>>
> >>> Spam Report:
> >>> Score     Matching Rule     Description 7.42     AWL     From:
> >>> address is in the auto white-list -2.60     BAYES_00     Bayesian
> >>> spam probability is 0 to 1%
> >>>
> >>>
> >>
> >> Sure, I've seen it do that lots of times. Is there a specific
> >> problem, or were
> >> you merely confused about the AWL having a large positive score?
> >>
> >> In general, you might want to read up on how the AWL works. Once
> >> you have a basic handle on what the AWL *really* is (hint: it's a
> >> score averager, not a whitelist), you'll have a better idea what to
> >> look at to try to figure out problems you may have with it.
> >>
> >> http://wiki.apache.org/spamassassin/AutoWhitelist
> >
> > I know with SA versions before 3.x it was recommended to turn off
> > AWL. Since then I have played with this being on and off. It works
> > fine for a while until a nice spam message or two get hit with a
> > nice big negative score and it sails through. I was wondering if I
> > am just not giving the system enough time to sort it's self out or
> > if SA with MailScanner is better off with out AWL?
> 
> Anyone got any opinions on this? It would be nice to know what others
> are running with. There was some inconclusive debate when SA 3.0 was
> released, which from memory resulted in a 'We'll give it a go and
> see'. I just wondered what if any results could be drawn.
> 
> Drew
> 
> 
> -- 
> In line with our policy, this message has 
> been scanned for viruses and dangerous 
> content by MailScanner, and is believed to be clean.
> www.themarshalls.co.uk/policy
> 
> ------------------------ MailScanner list ------------------------ To
> unsubscribe, email jiscmail at jiscmail.ac.uk with the words: 'leave
> mailscanner' in the body of the email. Before posting, read the Wiki
> (http://wiki.mailscanner.info/) and the archives
> (http://www.jiscmail.ac.uk/lists/mailscanner.html).
> 
> Support MailScanner development - buy the book off the website!
> 


-- 
Anthony Peacock       
CHIME, Royal Free & University College Medical School
WWW:    http://www.chime.ucl.ac.uk/~rmhiajp/
"Computer  software  consists of  only  two  components: 
ones and zeros, in roughly equal proportions.   All that is
required is to sort them into the correct order.

------------------------ MailScanner list ------------------------
To unsubscribe, email jiscmail at jiscmail.ac.uk with the words:
'leave mailscanner' in the body of the email.
Before posting, read the Wiki (http://wiki.mailscanner.info/) and
the archives (http://www.jiscmail.ac.uk/lists/mailscanner.html).

Support MailScanner development - buy the book off the website!