http://bugzilla.spamassassin.org/show_bug.cgi?id=3163





------- Additional Comments From [EMAIL PROTECTED]  2004-03-12 09:07 -------
With the patch above obfuscation is not suspected if the run of
non-whitespace after the tag does not consist of an alphabetic
character.  I've since gotten legitimate E-mail in which the run of
text before the tag consists of only non-whitespace, non-alphabetic
characters (a single open parenthesis).  With the patch below
obfuscation will be suspected only if the runs both before and after
the tag contain at least one alphabetic character.  

This (<xx>an let some true 0<yy>bfuscation through, but I'm guessing
that the limited number remaining obfuscation opportunities would
increase the effectiveness of the keyword scanners and Bayesian
classifiers that obfuscation is trying to evade.

As it stands (see below) the 0-10% obfuscation rule can't be scored
very highly, so the change should help.  (So would a 0-5 rule.)

>From http://www.pathname.com/~corpus/DETAILS.1day  (12 March 2004)
OVERALL%   SPAM%     HAM%     S/O    RANK   SCORE  NAME
 14.467  18.3443   3.3682    0.845   0.59    1.00  HTML_OBFUSCATE_00_10





------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

Reply via email to