[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Zimbra+SPAM=...?



Hi everybody,

we're trying to figure out the most efficient way of dealing with SPAM in Zimbra 
(we have edge MTA handling some of it too). What we're trying to find out what 
are the common practices in HiEd institutions? I'll bring some examples:

* do you let users "train" your Bayesian by marking mail as "Junk"?
* do you flush your Bayesian DBs every once in a while (and how often)?
* do you solely rely on Edge MTA for SPAM filtering and do only pass-through in 
Zimbra?
* etc. etc. etc.

We think our SPAM-filtering setup at present can't keep up with SPAM and we 
have either lots of false-positives or the contrary - lots of false-negatives.

I've seen some opinions that you shouldn't double-filter with Bayesian as your 
second instance would be "starving" and will not perform as expected. I also 
have met with opinions that you shouldn't do rule-based filtering and rely 
solely on Bayesian. Another opinion was that Bayesian is strictly a personal 
tool and you can't apply one SPAM/HAM DB across institution because 
preferences of HR clerk are different from Prof's and yet different from IT 
professional due to the different vocabularies of incoming mail. And so on and 
so forth. 

-- 
Dmitry Makovey
Web Systems Administrator
Athabasca University
(780) 675-6245

Attachment: signature.asc
Description: This is a digitally signed message part.