Once Mailtraq's anti-spam system has been set up in the Console it is recommended that day to day training is done using the normal Webmail client. Read more here.
The Anti-Spam Tab appears on the Mailbox Properties dialog and is used to configure the Bayesian Anti-Spam mechanism that will be applied to mail arriving in this mailbox. To read more about the Anti-Spam system, visit the Bayesian Anti-Spam topic.
The Database is a file containing all the features that have been learnt as well as a whitelist of senders whose mail will never be considered spam.
Each mailbox that enables the Anti-Spam system can either use its own database or share a named database. Each database is approximately 64mb, so it is probably unfeasible to have a separate database for every user in a system with many thousands of users.
Select Do no Analyze Mail for this Mailbox to disable the Anti-Spam system
Select Use the System Profile to have all Anti-Spam settings (including the database and actions to take when spam is detected) controlled by the System Profile (reached from Options | Bayesian Anti-Spam...)
Select Use the default database for this Mailbox to create and use a database named after the mailslot.
Select Use a shared database to create or use a database with the given name, which may also be used in other mailboxes. Any learning that takes place on this database will affect all mailboxes that use it. Note that other settings (such as the actions to take when spam is detected) will be controlled separately for each mailbox that shares this database
If a database is in use, a box will appear at the bottom of this tab giving information about the database status. Until at least 50 spam and 50 non-spam messages have been learnt, the database may not give reliable results. However, the Bayesian analyzer is cautious and until properly trained is more likely to favour false-negatives. Click on the More link for information on what the database contains.
Analyze on Receipt Tab
This tab allows the Anti-Spam system to operate on mail as it is being received (by SMTP). If the system determines that the message is spam, it can be rejected during the SMTP transaction and it will be returned to the sender with the given message. Once you are comfortable with the effectiveness of the Anti-Spam method, this is the recommended process as senders will be made immediately aware that their message has not been delivered.
The Spam Classification Point determines what score the message must be given before it is considered spam. You can have a different classification point to that used in Analyze in Mailbox, possibly being more cautious (though we recommend using the same point).
Redirect to address will result in the message being sent to an a specific address instead of it's original destination. You can redirect a message even if it is rejected.
Reject at SMTP with this message will return a permanent SMTP failure code with the text that is given here. You can use the macros $header(field) to refer to header fields and $hamtoken to return a unique address that can be used to bypass the anti-spam system for seven days.
Analyze in Mailbox
This tab allows the Anti-Spam system to operate as mail is stored in this mailbox. At this point the message content can be modified and the message can be moved to a specific folder.
Move Spam to... will place the message in the specified folder.
Modify Headers will add or modify a header in the message. This may be useful if your mail client can filter or move messages based on content. In the example below, the subject "Make Money Fast!!" might be replaced with "[SPAM:0.95] Make Money Fast!!".
The options on this tab allow the system to self-train based on the messages either sent to or received from this mailbox.
All mail to this mailbox is spam... should be selected if this mailbox is a Honeypot. A honeypot is an e-mail address made available in public locations (such as web sites and the Usenet) for the sole purpose of being found by address harvesters. If the mailbox only receives spam, it can be a useful early warning as mail can be learnt as spam without user intervention. Note: messages that can already be identified as spam will not be learnt (to keep the database optimised).
Mail with a return-path matching this mailbox will be learnt as non-spam should be checked if the mailbox represents the address of a local user who sends regular e-mail messages. Presumably the mail that the user sends out will be non-spam, and learning the messages as such provides another method for training the database without user intervention. Note: messages that can already be identified as non-spam will not be learnt (to keep the database optimised).
White-list all message recipients should be checked if you wish to have the recipients of all mail sent from this mailbox (i.e. messages with a return-path matching this mailbox) white listed. Normally this option would be selected in conjunction with the above option. The effective result is that if the owner e-mails someone, returning mail from them will not be analyzed.
This tab contains a manual list of addresses from which mail should not be analyzed. This list can be edited and may include wildcard characters (see Wildcards for more information). Do not confuse this white list with the automatic white list included within the database. That list can be edited by clicking on the More link on the Database Tab, and is shared by all users of the database file.
Each address (or wildcard specification) should be on a separate line.