Hi,
I'm noticing a lot of base64 encoded spam emails getting through ASSP 
for a while (Louis Vuitton & Rayban spams). These mails are being 
reported back to ASSP as errors, so there should be plenty of data 
available for Bayesian checking to catch them. So many of them are 
coming in that all spamtrap addresses are getting them, and some are 
getting removed from the corpus because of too many subject line 
duplicates. Analyzing a few of the mails with the Mail Analyzer 
indicates that the Bayesian score is near 1, which should end up 
spamming many of these emails. However, I notice in the logs that 
Bayesian checking is - mostly - not happening for these emails for 
regular users, and so they are slipping through with too low a score to 
be spammed. A Bayesian check did happen for a redlisted user and the 
mail was spammed as expected.

But, for some other (non-redlisted) users, a Bayesian check is indeed 
happening for these mails, but it is getting an extremely low score, 
such as 0.00000, which allows the mail through as the other checks do 
not bring the score up high enough. This leads me to think that perhaps 
the mail content is not being decoded properly on accept since there are 
plenty of these emails reported as errors, and collected by spamtrap 
addresses. The content is very similar and they do not have a bunch of 
random paragraphs from news stories and such in them in an attempt to 
throw off the Bayesian checking.

I am using ASSP 2.4.3(14258) (problem originally noted on b14241) on 
Perl 5.14.3

Is anyone else having similar issues with these spams?

-C

------------------------------------------------------------------------------
Slashdot TV.  Video for Nerds.  Stuff that Matters.
http://pubads.g.doubleclick.net/gampad/clk?id=160591471&iu=/4140/ostg.clktrk
_______________________________________________
Assp-test mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/assp-test

Reply via email to