Shouldn't a message that is identified as spam by the bayesian
filter of spamassassin (BAYES_90 or BAYES_99 in my case) never be
used as a message that is learned as ham?  (I would expect it 
not to be used for learning because it wouldn't improve the
bayesfilter, and training it as ham makes the bayesian filter
perform worse in future). Am I missing something?


On Fri, Jan 30, 2004 at 07:57:33AM -0500, Matt Kettler wrote:
> At 01:08 PM 1/29/04 +0100, PieterB wrote:
> >Can somebody explain me why a spammessage gets learned as ham?
> >
> >X-Spam-Status: Yes, hits=5.3 required=5.0 tests=BAYES_99,HTML_MESSAGE
> >        autolearn=ham version=2.60
> The score used for autolearning is the score as if bayes was not enabled. 
> Different scoreset, and no hits on any BAYES_ rules.
> Dropping BAYES_99, you're left with HTML_MESSAGE as your only rule hit.
> If you have network tests enabled, the score of HTML_MESSAGE with no bayes 
> and network checks on is 0.001.
> The autolearn threshold defaults to 0.1. This is well below that.
The crucial memorandum will be snared in the out-basket by
the paper clip of the overlying memo and go to file.

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