Daniel Quinlan wrote:

DQ> Craig R Hughes writes:
DQ>
DQ> > The right way is actually to have the AWL form a prediction based on a
DQ> > more sophisticated predictive model, including a zero-frequency
DQ> > estimate for senders who are not in the whitelist.  The weight
DQ> > provided by the AWL in cases where there is a-priori data about a
DQ> > sender should depend on the number of messages sent so far, instead of
DQ> > using the a-priori mean as 50% of the final score, the %age should
DQ> > shift over time.  There are well known ways of doing this optimally,
DQ> > but my reference book on the subject is up in Sausalito, and I'm down
DQ> > here in Menlo Park, 60-odd miles away.  I suppose I could go look
DQ> > things up online...  If only I hadn't smoked all that pot as a
DQ> > youngster, my memory might be good enough now to do it w/out reference
DQ> > books :) Or if I were less mathematically lazy, I could probably
DQ> > derive it from stuff I do remember.
DQ>
DQ> Does this also factor in the total number of senders seen in addition to
DQ> the number of messages received?  It seems like that's also a factor.

Yes, it would have to.

DQ> Also, it seems like we're lacking a good way to whitelist addresses
DQ> found in our own outgoing email in addition to incoming email.  One
DQ> could Cc: oneself and have procmail pipe to "spamassassin -W" option,
DQ> but I only want messages in the To: or Cc: to be added.

Well, the trouble is that SA isn't on the outbound path, and it's often a lot
more difficult to insert ourselves into the outbound stream than to insert into
the inbound stream.  What if you're reading your office email from your home
computer, and sending via your ISP's smtp server?  Just too many variables,
unless you have really good control over all your mail users.

C


_______________________________________________________________

Don't miss the 2002 Sprint PCS Application Developer's Conference
August 25-28 in Las Vegas -- http://devcon.sprintpcs.com/adp/index.cfm

_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk

Reply via email to