Forums & Blog

Hi,

how is the bayesian training done?
can i  manually start this process?

or is it done automatically each time i mark an email with spam or not spam?

thanks

It is done each time you mark mail as Spam. As far as I know marking a spam message as not-spam does not un-train the filters, it just adds it to the trusted senders list.

Check the "Messages required for filter update" entry in the Security -> Anti-Spam Administration -> Bayesian Filtering menu. By default the number is much higher than us "impatient" types would like :)

This will work. You have to create this two folders manualy, just put this in a *.bat and run it!!

cat X:\path to folder\spam|ham\fromail | spamc.exe -d 127.0.0.1 -p 783 -L spam

ramiss:

It is done each time you mark mail as Spam. As far as I know marking a spam message as not-spam does not un-train the filters, it just adds it to the trusted senders list.

Check the "Messages required for filter update" entry in the Security -> Anti-Spam Administration -> Bayesian Filtering menu. By default the number is much higher than us "impatient" types would like :)

Hi ramiss - what sort of lower level would you suggest? (I've just tried reducing it to 300).

It depends on how impatient you are versus how accurate you want it to be.  SmarterTools claims that it will not be accurate unless at least 1,500 emails have trained SA.  However, we set it at 100 in the beginning and it was very accurate from the start.