To protect against unsolicited email ( a spam ) the built-in mail client filter can be used. However even more effective solution is the filter placed on a mail server.
There are two ways of the filter usage ("Spam filter mode" field): immediate spam removal and message marking with special headers. The latest variant is the most preferable. On the one hand, this saves machine resources (by not starting a local filter), on the other, each user can decide himself what is considered as spam and how it is processed.
The filter analyzes a letter on many criteria and set it a final score. The higher it is, the more probability that letter is a spam. "Required score" parameter defines starting value, from which the letter will be recognized as a spam and, for example, removed. Letter evaluation criteria can be set independently with the name of the rule and it's score.
"Score" is calculated from one or four numbers.
RULE_NAME 0.5
RULE_NAME 0.5 0.6 0.5 0.5
RULE_NAME (0.2) (0.3) (0.2) (0.2)
One number means final rule score.
If four numbers are specified, one of them is choosed depending on use or
non-use of a statistical (on the Bayes algorithm basis) and network analyzers.
| statistical | |||
| none | use | ||
| network | none | 1st | 3rd |
| use | 2nd | 4th | |
Senders whitelist allows to estimate their reliability in long-term prospect.