The plug-in starts filtering spam immediately after the installation. But the email client performs filtering, whereas the plug-in only assigns a spam ratio to incoming messages. Therefore, you need to configure TheBat! first.

Select Options->Preferences->Anti-spam. The following options are available:

• Of 100 score to be used for actions below
  If you have only one plug-in installed, this parameter is useless.
• Delete a message if the score is greater than.
  Defines the minimum number of points needed before a message should be deleted. This is a very destructive parameter so don't enable it. You will be unable to recover misclassified and deleted emails otherwise.
• Move a message to the Junk folder if the score is greater than.
  Defines the minimum number of points needed to automatically move a message to the Junk folder. The higher value you set, the higher probability of false negatives you get (when spam is misclassified as normal messages). On the other hand, the probability of false positives becomes lower in that case (when normal messages are misclassified as spam). The default value is 60 and it is good enough in most cases. If you set it to 80-90, you will probably get more misclassified spam in your mailbox, but the probability of false positives will be much lower.
• Mark stored junk mail as read.
  This is a self-explainatory option and it is convenient to have it enabled.
• Move messages marked as junk to the Junk folder.
  If this option is enabled, messages will be moved to the Junk folder automatically when you mark them as spam.
• Use common Junk folder.
  You can enable this option if you want to. It doesn't influence the filtering quality.

Now you can start configuring the plug-in. Select AntispamSniper from the list and click Configure. The plug-in configuration window will appear. You can leave the default values or read the other topics of this manual and do something interesting ( Statistics , Filtering , Options ).

If you have some normal and spam messages in your mail client, it makes sense to train the plug-in using them. To train the plug-in, you can use the corresponding menu items in TheBat! - "Specials->Mark as Junk" and "Specials->Mark as NOT Junk". Just select one or more messages of either class and click the corresponding menu item. For your convenience you can specify shortcuts for these actions. New versions of TheBat! allow you to do it it in the Customize dialog box (View->Toolbars->Customize).

The whitelist is filled automatically when you are training the filter using non-spam messages. The more non-spam messages you mark, the more contacts will be listed. Messages from friendly addresses are not filtered and it is the best protection against being misclassified by other methods. The classifier needs messages of both classes. So feed it also with your junk messages if you have any.

The anti-phishing filter is enabled by default, but you can disable it if you don't need it. URIBL filter also blocks phishing mails along with spam, containing the links to blacklisted sites.

Tune up the DNSBL and Black words mechanisms in order to filter messages by their headers. To allow the plug-in to delete spam by message headers on the server and enable filtering IMAP accounts, fill up the list of accounts on the Servers tab. Don't forget to increase server timeouts in the account properties dialog box of The Bat! in order to avoid disconnections while the plug-in is checking message headers on the server.

Then you can enable the Testing mode and take a look at the Filtering log to see how the filter classifies existing messages. If you see any mistakes, you can train the plug-in using misclassified messages. The initial configuration is complete if you are satisfied with the results.

You can continue using the mail client as usual. It is necessary to review the Junk folder and search for misclassified messages. It is especially important in the beginning when the plug-in is not trained enough to avoid false positives. The probability of such mistakes will get lower fast if you train the plug-in.