RE: [IMail Forum] Statistical Filtering Settings?

[Jim Comerford]

Title: Message

It works best if every user sends it an equal number of "good" and "spam" mail.  To read up on it search for "antispamseeder".    Basically you need to create a MBX file for "GOOD" mail and one for "SPAM"... then you need to run the antispamseeder on those files to  "train" it.  We use the default settings and regularly "train" with new spam after an initial "training" with good mail.   Depending upon your user base you may need to "train" it more.  For example we have foreign language clients which we had to disable it for due to high false positives.  We also have a client involved with mortgages who needed to "train" theirs with lots of good mail to avoid false positives.   -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Dave Hauss Sent: Monday, March 22, 2004 8:56 AM To: [EMAIL PROTECTED] Subject: [IMail Forum] Statistical Filtering Settings? I sent a message the other day about the Statistical Filtering Settings and what is the best way to set this up.  Someone said I have to train it for every user. I don’t see how that is done in email. That would be a look of work for over 300+ users. I set it up for the following settings in IMAIL ADMIN:   Probability new word is spam 80% An Email is spam when its calculated prob exceeds 85% Max # of works to use 15   I am using the files for phrases and words etc that come from ipswitch.  Are these the best settings and if not what do U need to do? It’s catching some spam and letting others through.   Please email me back with any suggestions at [EMAIL PROTECTED]  ______ Dave Hauss www.pinehill.k12.nj.us Pine Hill School District MIS Building 1202 Turnerville Road Pine Hill NJ 08021 (856) 719-2103 (856) 719-2111 [Fax]