snip...

> That was our plan, keep the scores low, there are so many rules, you are
> likely to see an average of  5-15 hit on a spam message.  This is still
> low
> but it's safe for testing.  We've done lots of mass-checks against english
> ham and spam but we do not have access to any other language corpuses.  If
> you are willing to test and provide results back to myself or this group,
> we
> will appreciate it greatly.
>
> Even in a horrible senario where a ham message hit 50 of the tripwire
> rules,
> they would only produce a score of 3.85 well below the default threshold.
>
> It's common to see spam with 20-60 tripwire hits.
>
> If you know of a common abbreviation of a word for example GMC and you see
> it hitting a tripwire rule, please send me the details so I can fix up the
> rules.
>
> Thanks all, I hope they work great for everyone and at the time I am
> writing
> this, I don't see anymore spam with this stuff in it, now it's all
> high-level bayes busting words..
>


I installed them early this morning and they caught one of those spams
loaded with gibberish.  I had rule hits for a couple of screens.  If I
remember, each was for 0.08 but the total score was around 24.  There were
some other hits too but your rules made up the majority of the score.

Thanks for all the good work!


Gerry


-------------------------------------------------------
This SF.net email is sponsored by: Perforce Software.
Perforce is the Fast Software Configuration Management System offering
advanced branching capabilities and atomic changes on 50+ platforms.
Free Eval! http://www.perforce.com/perforce/loadprog.html
_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk

Reply via email to