On an external mail server, set up a training environment where misclassified mail can be filed. You must also configure a script to collect the mail from this training environment.
Setting up the Environment
The training environment consists of an IMAP mail server and email clients, which can directly access the mail server’s folder structure (like Microsoft Outlook, Mozilla, Evolution, etc). On the mail server, create the following mailboxes:
- SPAM – Contains spam that was delivered and not tagged.
- HAM – Contains mail that was incorrectly tagged as spam.
- FORGET – Contains mail that should not be tagged as spam or ham.
Users of the training mail server can sort misclassified into these mailboxes. SpamAssassin regularly collects and processes the mail from these mailboxes to improve its filter mechanisms. Depending on how accurately users can sort misclassified mail, you can set up these mailboxes either for all mail server users or for each mail user separately.
Environment for UNRELIABLE Users
For unreliable users, set up an environment where all users file misclassified mail into a shared HAM-SPAM-FORGET folder structure, which is detached from the training environment. The mail server administrator verifies that emails have been correctly sorted before moving them to the training environment. This approach may be provide additional work for the administrator but can help prevent the spam filter from learning incorrect mail classification.
Environment for RELIABLE Users
For reliable users, set up an environment where all users have access to the training environment on the mail server and file misclassified mail into the corresponding mailboxes.
Each user has the HAM-SPAM-FORGET folder structure and sorts misclassified mail into these folders.
Configuring a Script for Email Collection
You must configure a script to collect emails from the training environment.
For example, you can use a script that is similar to the following:
#!/bin/bash # assumptions: # HAM and SPAM live under /home/$USER/mail/ # TARGETDIR should not be /tmp/, but a more secure location # no filelocking, etc # 2003-12-18 email@example.com TARGETDIR=/tmp/ SPAM=`find /home/*/mail/ -type f -name SPAM` HAM=`find /home/*/mail/ -type f -name HAM` for a in SPAM HAM; do if [ -f $TARGETDIR/$a ]; then rm -f $TARGETDIR/$a fi done for a in $SPAM; do cat $a >> $TARGETDIR/SPAM done for a in $HAM; do cat $a >> $TARGETDIR/HAM done
Configuring a Training Environment on a German MS Exchange Server
If your training environment is configured on a German Microsoft Exchange Server, specify the paths to the mailboxes as follows:
Mailbox SPAM:&ANY-ffentliche Ordner/spamassassin/SPAM
Mailbox HAM:&ANY-ffentliche Ordner/spamassassin/HAM
Mailbox FORGET:&ANY-ffentliche Ordner/spamassassin/FORGET
After setting up the training environment, configure the training settings on the Barracuda NextGen Firewall F-Series. Continue with How to Improve Spam Filtering.