Beyesian Filter - Good dictionary and acceptable levels

UPSdata · Post by **UPSdata** » Mon Oct 09, 2006 9:01 pm

I am in the process of upgrading from PRO to ENT version of ME.

Playing with the beyesian filtering and wondering if anyone has a good dictionary set, or if there is a good resource for pulling a community maintained one?

Also, what is a good level to set for filtering the spam, but not getting false positives. Right now I have it set at 95%.

TIA

-Erik

rockinthesixstring · Post by **rockinthesixstring** » Tue Oct 10, 2006 5:15 am

everyones dictionary will be different and you should not use someone else's. your best bet is to get your hands on the Bayesian Whitepaper provided by MailEnable Ben and configure your own.

lschafroth · Post by **lschafroth** » Wed Oct 11, 2006 2:34 pm

I struggled with that until I got the Whitepaper from Ben.

I was able to build my own custom dictionary and create scripts that poll the spam folders every 2 hours and update the dictionary.

We went from 10-20 spams a day getting through XBL, Reverse DNS and etc, to maybe 1 a day or less!

When I do get a new one that gets through I just move it to my SPAM folder and 2 hours later it's part of the dictionary.

Lannie

rockinthesixstring · Post by **rockinthesixstring** » Wed Oct 11, 2006 4:16 pm

yep, thats exactly how it works. mine does the same thing.