----- Original Message -----
From: "jdow" <[EMAIL PROTECTED]>
To: <users@spamassassin.apache.org>
Sent: Saturday, August 26, 2006 9:51 PM
Subject: Re: Pse help stock spam
From: "John Fleming" <[EMAIL PROTECTED]>
My server is getting more and more "stock spam" with an image and a few
paragraphs of bayes-obfuscating text. I've been chasing them by either
learning them or increasing the score of SARE or other rules that hit.
Today I got several that didn't hit anything useful and were learned as
ham:
X-Spam-Status: No, score=0.0 required=5.0 tests=BAYES_50,HTML_30_40,
HTML_MESSAGE autolearn=ham version=3.0.3
What is the consensus advice regarding how to stop these? Should I still
be learning them even with the obfuscating text? Seems like that would
be creating tokens against lots of normal words??
FuzzyOcr helps. Search for it on the spamassassin wiki pages as a
plugin. It's still in development, though.
Another tool that helps some is the ImageInfo plugin that Dallas did.
I'm not sure where its current incarnation exists.
{^_^}
Thanks jdow and Mike! All I needed for now was the SARE stock rules.
Somehow I had missed those! The trash all went away with the SARE stock
rules addition, and I find them all in the spam bucket.
Thanks SA and developers/supporters! - John
