On Friday, February 24, 2006, 2:56:02 PM, Andy wrote:

AS> Hi,

AS> I'm realizing that some Sniffer rules amount to nothing more than IP
AS> blacklists.  

AS>         received:~+[nnn\.nnn\.nnn\.nnn]
AS>                         
AS> Are all "sender IP" rules properly grouped so that I can identify and ignore
AS> them by return code. I already use IP blacklists (and other means) to "cross
AS> check" Sniffer and add to my "confidence" value before a mail is finally
AS> blocked.

AS> I can't afford Sniffer to effectively "double up" those sender-IP tests.
AS> Ideally, Sniffer should perform content checking.

Please review the result code explanations here:

http://www.sortmonster.com/MessageSniffer/Help/ResultCodesHelp.html

IP rules are coded to symbol 63. The voting system on each SNF node
sees rules with lower symbol values as "more fit", so the only time
you will see a result code of 63 is when no other rule matches that
message.

You may want to reconsider ignoring this result code - there is added
value.

When an IP rule is in the SNF rulebase, it indicates that:

* The rule is from a message that reached our spamtraps.

* Additional algorithms were used to classify the IP as a spam source.

* The source has been consistently and recently active and detected at
our user's system. Inactive IP rules are "forgotten" after a short
period.

* There have been no false positives reported against the rule. We
remove IP rules on the first FP case and place the rule in a
"problematic" rule group so that it cannot be reinstated without a
strict review.

* No other rules in our system are currently identifying the
associated message content. Though we do focus on content, it is clear
that in some cases an IP is the most efficient indicator.

Since most other blacklisting services focus on a broad spectrum of
IPs, there is bound to be overlap between them and also with SNF IP
rules. However the fact that the IP shows up in our system does carry
some unique information about that IP (see above).

We explicitly do not aggregate IP rules from other lists. We recognize
that other IP black lists are used in spam filters along with SNF and
we encourage that as well as the use of other tests. (Even though SNF
encapsulates diversity in it's algorithms and continues to expand this
diversity, the best filtering systems will always use as many useful
mechanisms as possible.)

Additionally, as we move forward, IP rules in the SNF ruelbase will be
gathered by unique, sophisticated mechanisms such as wavefront
detection and cross-feature source correlation, etc. As a result, IP
rules found in the SNF rulebase will increasingly represent some
unique characteristics not found in other IP lists.

Hope this helps,

_M



This E-Mail came from the Message Sniffer mailing list. For information and 
(un)subscription instructions go to 
http://www.sortmonster.com/MessageSniffer/Help/Help.html

Reply via email to