Re: my solution to image spam (previously mentioend consistant failures)

TBB suggested adding tokens from the mime headers to the mix as a way of 
maybe getting more useful tokens.  I have not tried this myself but 
thought I'd pass it along for anyone who wants to.  You might try 
something like this:

  spamprobe -o suspicious-tags -H+content-transfer-encoding \
     -H+content-type -H+content-id train

(note the \ to wrap in the right place - really it's one command line)

Not sure if the headers will help or not but they might be useful.  My 
intuition is that the tokens from these headers are likely to be pretty 
neutral if spammers are smart enough to randomize file names and 
attachment ids.

On the subject of how spammy inline images really are: in my database 
terms containing cid have the following data:

0.8337588     388    8854       1  U_cid
0.9721581      31    4925       1  Timg_cid
0.9721361      31    4921       1  Timg_src cid

So it seems that cid is a good spam indicator but not a perfect one for 
my own email.

All the best,
++Brian

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

_______________________________________________
Spamprobe-users mailing list
Spamprobe-users-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
https://lists.sourceforge.net/lists/listinfo/spamprobe-users