[spambayes-dev] interesting spam

[Skip Montanaro]

Attached is an interesting spam - well, I found it interesting.  Aside from
all the obscure words apparently designed to lower the score, it appears the
message was mostly one big line.  Someone's input buffer broke it in a few
places before the randomizer could replace $RANDOMIZE with a token "good"
word.  There are even token words in the URLs (boldness and tenebrous).

The only potentially interesting tokenizing gimmick I see is the ratio of
HTML tags (or runs of tags - "<br><br><br>" would be one, not three) to
words (very high here).  I don't know if that's a decent spam discriminator
or not, but it's kind of hard to imagine a person writing a non-spam email
of that length with that many <font> tags.

Skip

-------------- next part --------------
An embedded message was scrubbed...
From: "Tayeb Homec"<itine at YAHOO.COM>
Subject: exceedingly
Date: Thu, 31 Jul 2003 23:47:37 GMT
Size: 8021
Url: http://mail.python.org/pipermail/spambayes-dev/attachments/20030731/e3d81035/attachment.eml