Success !!!! Re: spamprobe-1.3x2 released

"David A. Lee" <[email protected]> Mon, 5 Dec 2005 14:14:44 -0500
Newsgroups gmane.mail.spam.spamprobe.general
Message-ID <[email protected]>
I located and installed the libungif and the new spamprobe. Ran a smoke test 
(nothing exploded) and installed to production ("production" means 3 users 
:)

No problems at all, and great success !
First off I didnt want to have to train from scratch on my annoying GIF 
spams so I ran

spamprobe train-spam  <my archive of spam>

I dont know if this did any good because <my archive of spam> was already 
tagged as spam ! but oh well gave it a shot.

one day later I am NO LONGER getting ANY of those annoying stock market 
spams with only GIFS. These have been my bane for several weeks now, but now 
they are gone.

I am a bit concerned that the GIF headers being used as the tokens are 
primarily the width and height ... but hey, so far I havent been hurt by 
spamprobe taking other things I think are "not spammy" and treating them as 
important spam tokens so I'll just trust ... so far so good (but I rarely 
get any non-spam with just a GIF ... so its not a fair test)

Good though ! Here's an example of -T output on one of these stock market 
spams ...
X-SpamProbe: SPAM 0.8550869 a2efe007d7e0382fc05a7b6ae1c8e0e3
    Spam Prob   Count    Good    Spam  Word
    0.9984227       1       0       5  4bdd2d2b3c366c9eb8551b9814b03561
    0.9984227       1       0       5 
Igif_digest_211ee9ede7ede43ad72fb704667037d1
    0.9984227       1       0       5  U_4bdd2d2b3c366c9eb8551b9814b03561
    0.9984227       1       0       5  U_cid 
4bdd2d2b3c366c9eb8551b9814b03561
    0.9984227       1       0       5  cid 4bdd2d2b3c366c9eb8551b9814b03561
    0.9803216       1     105   20340  Hto_david calldei.com
    0.9594344       1     191   17566  Hrecv_david calldei.com
    0.9593478       1     191   17527  Hrecv_for david
    0.9590801       1     195   17772  Hrecv_david
    0.0986668       1    2408    1025  Hrecvx_dec
    0.8437633       1       3      63  Igif_image_0_width_480
    0.7996497       1      25     388  Hrecvx_smtp mon
    0.2200519       1     103     113  Hrecvx_mon dec
    0.7711100       1      10     131  Igif_image_0_left_0
    0.7711100       1      10     131  Igif_image_0_noninterlaced
    0.7711100       1      10     131  Igif_image_0_top_0
    0.7669953       1      10     128  Igif_image_0_no_color_map
    0.7627844       1    1332   16655  U_cid
    0.2391936       1    1474    1802  Hrecvx_from unknown
    0.2451723       1    1730    2185  Hrecvx_unknown
    0.2462083       1    8501   10797  Hrecvx_mon
    0.7319334       1    1847   19610  cid
    0.2879835       2   39256   31948  Hrecvx_IP_ADDRESS
    0.7059032       1       3      28  Igif_image_0_height_640
    0.7045590       1    8518  117933  Hrecv_smtp id
    0.3000000       1       0       0  Hfrom_directstock





> I've just released another experimental spamprobe snapshot.  This one
> adds a feature suggested by Nico to help combat wordless spams.  If
> configure finds a working libungif and gif_lib.h on your system then
> spamprobe will use that library to extract useful terms from gif
> attachments in emails.
>



-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
for problems?  Stop!  Download the new AJAX search engine that makes
searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
http://ads.osdn.com/?ad_id=7637&alloc_id=16865&op=click