The filtering window contains all information about heurastic scanning. The complete mailmessage will be tested on multiple items, including bayesian analysis. The definable threshold level determines if a message is spam or not.
All these tests together result in a total score. The higher the score the more likely it is that the message is spam. It’s also possible to modify the subject when the score exceeds a certain level.
The Enable heurastic filtering checkbox will enable the heurastic scanning of mail messages and enable the concerning panel when checked.
Option avalailable are:
Forward to mailbox
If checked all heurastically detected spam messages will be forwarded to another mailbox.
The definition of the forwarded mailbox is set here. The original user is referred to as #USER# and the original domain is referred to as #DOMAIN#.
In case of the emailadres ‘firstname.lastname@example.org’ the variable #USER# will be ‘test’ and the variable #DOMAIN# will be ‘acme.com’
In case of the filled in example, spammail will be forwarded to ‘email@example.com’.
Modify subject if spam
If checked this setting will enable the inputbox Modify subject suspected spam. Pinjo can modify the subject if the score exceeds the value defined in Mark as spam threshold level
Modify subject suspected spam
This inputbox contains the new subject of the mail message in case the message will be marked as spam. The variable #SUBJECT# will be replaced by the original subject of the message.
Mark as spam threshold level
Here the level is specified at which the message must be marked as spam. If the determined score of the message exceeds this level, the message will be marked as spam. The default value is 5. This is a reasonable reliable score.
Drop message if score is higher as
This checkbox gives the possibility to drop the message at a certain score. Choose a score here where you are absolutely sure that the message will be spam. Default value is 15. Scores higher as 15 are most probably spam, as they have so much positive tests.
Skip mailmessages larger as
Since most spam messages or not particularly big, you can set the level when a message will be skipped for testing. Size is set in KiloBytes here.
Note: This skipping of messages ONLY involves the heurastic scanning. Viruschecking and others tests will always be done.
The Heurastic statistics panel shows details about the bayesian database. Statistics information is only updated on pressing the Statistics button.
The number of messages in the database considered as spam.
The number of messages in the database considered as non spam (also called ‘ham’).
The number of tokens in the database. A token is one word.
Date of the last token in the database. It’s possible that this value is a bit in the future.
Last sync time
Date and time of the last synchronisation of the database.
Date of the oldest tokens in the database. Tokens will not stay forever in the database. Older non used tokens will be removed from the database at sync times.
The Heurastic learning panel shows when and how the manual heurastic learning will be done.
This learning is an addition to the the automatic learning. If messages are not marked as spam, or most surely or not spam, you can place a copy of those messages in the appropiate folder (./spam for spam and ./ham for non-spam). Pressing the Learn now button will analyse the message and update the heurastic database for statistical analysis.
Messages can only be learned once. If a message is already learned before, it will be skipped.
Autolearn at midnight
When checked, mailmessages placed in the appropiate folders will be automatically processes at midnight.
Delete files after learning
When checked, files processed for learning are deleted after processing. This will involve automatic and manual learning.