>>Hello, I have used this program for a while, but it still marks some
legitimate mail as spam. Does anyone have a sample cfg file that blocks a
good amount but doesn't mark legit mail spam I could look at? Do most people
just have it mark it spam or have it automatically delete it? Thanks, Mark<<
Below is the GLOBAL.CFG file I use. I have edited out our whitelist entries
and custom blacklists. You'll notice a few things that are different from
the default file:
1) I am using the MAPS tests, which we pay a yearly subscription for. After
using it for a while, along with all the other free sites, I am not so
impressed with it. From reviewing the headers of spam to see which tests
caught it, I really don't recall where MAPS was the only one to catch it.
There are usually a couple of other tests that also catch the same spam. I
don't see any distinct advantage to MAPS at this point that is worth paying
for. Literally, I could probably drop it and not affect my spam catching
ability too much.
2) Where a blackhole site has separate return values for different tests, I
do not look for the separate return values - I just use the wildcard "*" to
accept any return value. I presently am not distinguishing which test from a
site provoked a positive response.

3) I have two WEIGHT tests. I use a WEIGHT12 test to decide whether or not a
particular email is spam. I use 12 because it has a number of factors,
allowing me to fine tune the relative weighting of the tests. Based on my
experience with the reliability of a test, those that I have high confidence
in get a weight of 12, allowing a single test in some cases to impeach an
email. Other tests that generally do a good job get a weight of 6, requiring
two of them to concur before considering an email spam. Finally, for those
tests that are less reliable, I give them a weight of 4, requiring three of
them to concur before mail is branded as spam. Note that I could extend this
to weights of 3 and 2, requiring concurrence of four and six tests
respectively. After reviewing the headers of many messages for the weights
they received, I noticed that some messages got tagged by so many tests as
being spam that they had weights in the stratosphere! I decided that with
that much concurrence among all the tests that I use, the chances of an
email being spam approach 100%. So, I have a weight test of WEIGHT53. The
value of 53 has been determined by basically trial and error of what value
defines a threshold above which there were zero false positives.

4) In my "$default$.junkmail" file, all tests are assigned the "LOG" action,
except for WEIGHT12, which gets "WARN" and WEIGHT53 which gets "HOLD". Our
use of IMail and Declude Junkmail is in a corporate environment. All of our
email users are standardized on the use of Outlook 2000. We use Outlook
rules to look for email headers that contain the string "X-RBL-Warning:" and
move messages with such headers to a "Junk Mail" folder. Our users are asked
to periodically review the messages placed there for false positives before
they delete the spam. Messages that get tagged by the WEIGHT53 test and are
put in the \IMail\spool\spam folder are reviewed by me before deletion. I am
looking at using the SpamReview program to assist me in this effort.
Although I am sorely tempted, I presently do NOT automatically delete any
mail, for fear of an important business communication receiving a false
positive. This may change - I may decide on a weight threshold high enough
that I would have the confidence to delete it without review.

Here's my GLOBAL.CFG. The formatting of the columns got somewhat skewed when
importing it into this email...

#
# Declude JunkMail configuration file
#

CODE            [Insert your code here]

# The "####" in the LOGFILE option gets replaced with the month/date with
v1.11 and higher

LOGFILE         D:\IMail\spool\dec####.log
LOGLEVEL        LOW
LOG_OK          NONE

# Hop is set to 4 because we outsource our virus scanning and have 4
mailservers in front of our mail server.
HOP             4
#HOPHIGH        1

#
# Below are some advanced options
#

CONSOLE OFF
XINHEADER       X-Note: JunkMail tests failed: %TESTSFAILED%
XINHEADER       X-Note: This E-mail was scanned by Declude JunkMail v%VERSION% for
spam.
XOUTHEADER      X-Note: This E-mail was scanned by Declude JunkMail
(www.declude.com) for spam.

#IPBYPASS       127.0.0.1
#XSENDER        OFF
XSPOOLNAME      ON
#XINHEADER      X-Note: This E-mail was sent from %REVDNS% ([%REMOTEIP%]).
#XOUTHEADER     Organization: Your Name Here

MAPS-RBLPLUS    ip4r    rbl-plus.mail-abuse.org           *             12      0
OSRELAY                     ip4r    relays.osirusoft.com               *
6       0
ORDB                        ip4r     relays.ordb.org                      *        6   
 0
SPAMCOP                   ip4r    bl.spamcop.net              127.0.0.2 12      0
ORBZIN                       ip4r    inputs.orbz.org              127.0.0.2
6       0
VISI                             ip4r    relays.visi.com
127.0.0.2       12      0
COMPU                       ip4r    blackhole.compu.net       127.0.0.4
6       0
COMPU-PM0               ip4r    pm0-no-more.compu.net  127.0.0.4       12
0
NJABL                         ip4r    dnsbl.njabl.org                 *
6       0
SUMMIT                       ip4r    blackholes.2mbit.com
127.0.0.2       12      0
WIREHUB-DNSBL         ip4r    blackholes.wirehub.net          127.0.0.2
6       0
WIREHUB-DYNA          ip4r    dynablock.wirehub.net           127.0.0.2
6       0
DORKS                      ip4r    orbs.dorkslayers.org           127.0.0.2     6      
 0
# Fiveten seems to return too many false positives
FIVETENSRC      ip4r    blackholes.five-ten-sg.com      *               4       0
GUARDBLOCK      ip4r    spamguard.leadmon.net           *               6       0
SPAMBAG         ip4r    blacklist.spambag.org   127.0.0.2       6       0
# Blars tagged mail from yahoo.com as spam!
BLARS                      ip4r block.blars.org                 *       6       0
BLITZED                 ip4r    opm.blitzed.org                         *              
 6       0
DEVNULL         ip4r    dev.null.dk             127.0.0.2       12      0
DEWS                         ip4r    dews.qmail.org
127.0.0.2       12      0
FABELSOURCES        ip4r    spamsources.fabel.dk         127.0.0.2      12      0
FLOWGO                    ip4r    flowgoaway.com                127.0.0.2       12     
 0
INTERSIL                  ip4r     blackholes.intersil.net      127.0.0.2       12     
 0
SELWERD           ip4r     xbl.selwerd.cx               127.0.0.2       12      0
MONKEYPROXIES      ip4r    proxies.relays.monkeys.com  127.0.0.1       12
0
MONKEYFORMMAIL   ip4r    formmail.relays.monkeys.com  127.0.0.1       6
0

DSN                      rhsbl  dsn.rfc-ignorant.org                     127.0.0.2     
 6
0
DNSUCE                       rhsbl     in.dnsbl.org
*               12      0
NOABUSE         rhsbl   abuse.rfc-ignorant.org                   127.0.0.4      4      
 0
NOPOSTMASTER    rhsbl   postmaster.rfc-ignorant.org              127.0.0.3      4      
 0
MONKEYSENDER        rhsbl   sender-domain.sjesl.monkeys.com     127.0.0.1
6       0
PIGS                            rhsbl   bandwidth-pigs.monkeys.com
127.0.0.1       6       0

MAILFROM                   envfrom              x       x       12      0
SPAMHEADERS     spamheaders     x       x       12      0
ROUTING         spamrouting     x       x       12      0
PERCENT         percent         x       x       12      0
BADHEADERS      badheaders      x       x       6       0
REVDNS          revdnsexists    x       x       4       0

WEIGHT12        weight          x       x       12      0
WEIGHT53        weight          x       x       53      0



---
[This E-mail was scanned for viruses by Declude Virus (http://www.declude.com)]

---

This E-mail came from the Declude.JunkMail mailing list.  To
unsubscribe, just send an E-mail to [EMAIL PROTECTED], and
type "unsubscribe Declude.JunkMail".  You can E-mail
[EMAIL PROTECTED] for assistance.  You can visit our web
site at http://www.declude.com .

Reply via email to