fyi - if i check the LearnAsSpam mbox the next day - it is empty, means it is getting deleted.
Then it is working as expected.
why i think the Learn contribs is not working : i keep on putting the same spam em ( eg newsletters@response.enterprise-alerts.com ) into the LearnAsSpam mailbox, but it is just not being spamtagged the next time. The box is running since quite some time and should have received by far more than 500 LearnAsSpam emails. I just have the feeling it is not learning ....
I think you will need to read up on how spam filtering works, especially on the topic of bayesian filtering. In short each message is fed to the LearnAsSpam script, it will learn it's tokens from the message. These tokens are then used by the bayesian filter to test your e-mail. If the bayesian filter considers it a positive it will add the score assigned to this test to the total spam score.
Normally the bayesian filter needs a minimum of 200 messages (IIRC) before it will even start testing on this filter.
In general as long as the total spam score is not above your spam treshold it will not be tagged as spam, since the bayesian filter implemented by the learn scripts is just a test it will not block spam, it will score it if the minimum number of messages required is reached, which should be visible in the mail headers as this test should then also be listed in the tag that holds the test ran in the header.
You can use the following command to judge if enough messages have bean learnt:
sa-learn --dump magicwhich should look something like this:
[root@sme74test ~]# sa-learn --dump magic
0.000 0 3 0 non-token data: bayes db version
0.000 0 1635 0 non-token data: nspam
0.000 0 25 0 non-token data: nham
0.000 0 132133 0 non-token data: ntokens
0.000 0 1163576731 0 non-token data: oldest atime
0.000 0 1239124533 0 non-token data: newest atime
0.000 0 0 0 non-token data: last journal sync atime
0.000 0 0 0 non-token data: last expiry atime
0.000 0 0 0 non-token data: last expire atime delta
0.000 0 0 0 non-token data: last expire reduction count
[root@sme74test ~]#The nspam entry lists the number of spam messages learned.
4 Replies
1153 Views