Re: Barracuda Spam Filter

[Bruce Hudson <Bruce.Hudson () DAL CA>]

> My apologies to participants on the list.  I will not respond again.

    I do not think you have anything to apologize for. We too are getting
about 90% of the spam that enters our site. Modern Bayesian spam filters
may be accurate in the high 9's when they are used with a small, coherent
group of recipients. At least in my experience, they do less well when you
try to put one in from of an entire university. A School of Pharmacy can
play havoc with detecting the drug spam.

    Individual users get to apply further filters, including their own
Bayesian filters, as the fetch their mail but we have 90% of the 80-85%
of the incoming mail that is the SPAM that our IMAP servers do not have
to deal with. While there is definitely room for improvement, I am not
quite ready to throw out what we have as worth-less.
--
Bruce A. Hudson                         | Bruce.Hudson () Dal CA
UCIS, Networks and Systems              |
Dalhousie University                    |
Halifax, Nova Scotia, Canada            | (902) 494-3405