This solution applies to all Barracuda Spam Firewalls, all firmware versions.
Administrators of the Barracuda Spam Firewall must to classify a minimum of 200 messages as Spam and Not Spam on the Basic > Message Log page, which will automatically update the Barracuda Bayesian Learning database (displayed on the Basic > Bayesian/Intent page, under the Barracuda Bayesian Learning heading).
When this has been done, the Bayesian database will begin to adjust the spam scores of incoming messages. It is also recommended to have several more messages classified as Not Spam than Spam.
For more help configuring your Bayesian database, see Solution #00001324.
Link to This Page: