I'm in the midst of my second attempt at getting the bayesian filter working as our spam filter. I used to just tell customers to install their own spam filter locally but lately more and more want it at the server level, or they're webmail only users. My previous attempt at training the bayesian filter was pretty unfruitful. Lots of false positives even after weeks of training.
So it's been a few years, new versions of ME have come out, i'm giving it another go. Being a little more aggressive in my manual training (going through ALL messages marked as spam and moving them to a ham folder to be trained, and going through all outgoing authenticated email looking for possible compromised accounts sending out spam). But it's exhausting...
I'm just curious if anyone else on here has found good results with the bayesian filter? I run one locally on my outlook and it works great, but that's for just me. I'm dealing with hundreds of different people on the server, so what qualifies as spam kind of varies.
Does the built in bayesian filter work sufficiently?
Re: Does the built in bayesian filter work sufficiently?
It isn't fantastic. I use http://www.mxuptime.com/
---
Shaun Rieman
Advanced Micro Technologies, LLC
Shaun Rieman
Advanced Micro Technologies, LLC