*grin* I imagine that particular confusion probably would come up a bit.

Yes, treating it like the virus definitions would be an appropriate step - available, but with solid warnings about usage.

I'd still really like to see something that deals with hoaxes and chain letters separately - a "sideways" scoring system that could avoid changing to overall spam score, but provide extra header info would be great.

Perhaps a few extra score fields and ways to add and subtract from them in SpamAssassin could do it... ie, SCORE_1 1.5 would add a score of 1.5 to a secondary score counter. You could have as many as you'd like, and be able to rate/sort/label e-mail based on a large number of things, not just spam.

Bayes, however, is ideally suited for hoaxes, and it would be great if we could include that in a hoax solution.

Evan II

-----Original Message-----
From: Chris Santerre
Sent: Thursday, May 20, 2004 3:14 PM
To: 'E. Falk'; [EMAIL PROTECTED]
Subject: RE: Scoring Hoaxes


LOL, don't take my post the wrong way. It seriously wasn't towards you. It hit an exposed nerve having to do with something else.

I understand your concerns about hoaxes. I believe they will get treated
like the virus rules:
http://www.exit0.us/index.php/VirusBounceRules
Which may be created in a cf file, but not in RDJ. And certainly with a big
HUGE warning label about FPs, bayes DB changes, and possibly ear infections.
:)

--Chris



Reply via email to