Hi,
I use mailwatch to browse through spam - I am using filters to find spam and after marking spam/ham I click "learn" button. I am doing this on a daily basis since months but I still have the same spam coming to my server and SA dont even raise score after many "learn" processes.
Since the beggining of MS installation I am getting "lets chat", "Can we talk?" and "enlarge your male organ" and those messages are scored 5 - 9.
My question is: How exactly learn proces works - does clicking on "learn" send spam for analyse and where it is analyzed.
best,
Piotr
MailWatch - marking msg as a spam learn actually SA?
-
nabuhonodozor
- Junior Member
- Posts: 48
- Joined: 29 Oct 2007, 07:01
The learn function just corrects the mail's inclusion in the Bayesian database as ham or spam, it won't change any of the other scores (besides bayes) any future identical spam will receive.
If you are getting a lot of false negatives, you might check to make sure that razor, DCC and bayes are all working on your server. You can search for these in MailWatch reports, use Spam Report contains and then either razor, dcc, or bayes (not all at once) to see if they are actually being used in the spam scoring. These are the types of tests that will quickly start catching "new" types of spam.
If you are getting a lot of false negatives, you might check to make sure that razor, DCC and bayes are all working on your server. You can search for these in MailWatch reports, use Spam Report contains and then either razor, dcc, or bayes (not all at once) to see if they are actually being used in the spam scoring. These are the types of tests that will quickly start catching "new" types of spam.