On 1/6/07, Raj Shekhar <[EMAIL PROTECTED]> wrote:
in infinite wisdom Eugen Leitl spoke thus On 01/04/2007 05:27 PM:
> On Thu, Jan 04, 2007 at 05:12:35PM +0530, Deepak Misra wrote:
>> I keep getting such mails. Are these some probes to test spam filters
or are
>> they to confuse them ??
>
> Bayesian spam filter poison. The payload is in the image.
Spamassassin has been able to tag most of these messages as spam
successfully for me(I am using debain's default setup). For example,
here is X-Spam-Status from one such message
"
X-Spam-Status: Yes, score=24.406 tagged_above=-1000 required=3.5
tests=[BAYES_99=3.5, DNS_FROM_RFC_WHOIS=1.447, EXTRA_MPART_TYPE=
1.091,
HELO_DYNAMIC_DHCP=3.066, HELO_DYNAMIC_IPADDR=4.2,
HTML_IMAGE_ONLY_04=3.6, HTML_MESSAGE=0.001,
RCVD_IN_BL_SPAMCOP_NET=1.558, RCVD_IN_SORBS_DUL=2.046,
RCVD_IN_XBL=3.897]
"
A large number of small scores add up to give such messages a high score.
--
raj shekhar
facts: http://rajshekhar.net | opinions: http://rajshekhar.net/blog
I dare do all that may become a man; Who dares do more is none.
I am curious: Is it possible to use this technology in reverse and use
these filters to try and guess the author of a mail given a collection of
previous works by many people ??
Deepak
