To get a good filtering accuracy it is necessary to train the Markovian
spam filter
Just do as demonstrated here.
First check your email with SpamRIP
The Markovian filter has not been trained, so each incoming email is
seen as a friend (first column is green)
BUT, the first emails come from Spamcop blacklisted addresses or blacklisted
countries (see above, the country filter), so SpamRIP automatically proposes
teaching the Markovian filter recognizing these emails as foe. |
 |
 |
Just accept (i.e. click on "Process messages") and SpamRIP
will :
- delete the selected messages directly on your remote mailboxes
and LEARN
- All other emails are non affected
|
 |
 |
Now, just "Clear all data" (red cross) and "Receive
all messages" again, SpamRIP will read again all your incoming messages
taking into account what he just learn.
The Markovian filter proposes (rightfully) to consider most incoming
messages as spam.
Change the status of those emails the various filters have no yet recognized
correctly, here an email is coming from a "friend" source but
is recognized as spam by the Markovian filter
The Markovian filter is right, so change the check-box accordingly (uncheck
"Change & lean", check "Delete") before processing
the emails. |
 |
 |
Process by clicking on "Process messages"
When processing messages the actions noted in checkboxes are performed.
Remember, it will take some time before the Markovian filter have learn
enough to be 99.9 % accurate. |
 |
| |
|
| |
|