Re: Evo, spamassassin, razor
"Scot L. Harris" <[email protected]>
| Newsgroups | gmane.comp.gnome.ximian.user |
|---|---|
| Message-ID | <1082937820.21367.8.camel@lathe> |
On Sun, 2004-04-25 at 12:40, Bob James wrote: > Finally getting fed up with the spam arriving in my multiple email > boxes, I googled and found this page on using spamassassin with Evo: > > http://www.atlantawebhost.com/articles/evolution_spamassassin.php > > > 1) Setting up a whitelist, and > 2) Easy reporting of new spam fingerprints. > > Is anyone doing this with their systems? Ideally, this would be a wholly > client-side operation. Any suggestions? Been doing this for some time. The whitelist is easy, in your home directory there should now be a .spamassassin directory. You may find a user_prefs file already there. If not you will need to create one. In that file you can put your white_list entries. As for 2) I think what you want is to train the bayesian database with unmatched spam. That is easy also. I created a subfolder for such messages. When I get one I just move it to that folder. Then all you need to do is run the sa-learn command and pass it the mbox folder with the unmatched spam. A command like the following: sa-learn --show-dots --spam --mbox ./mbox NOTE: in order for the bayesian database to be used you must have passed it 200 or more messages. The easiest way to do this is to collect spam for a few days and run the sa-learn command against them. You may have already done this. If not you will find the match rate greatly increases once you have done this. Also be sure to run ham messages through the database also. Same command but use --ham instead. Once you are comfortable with the process you can automate it easily. I have not bothered yet but a script that runs via cron each night or once a week would be easy to write. Hope that helps. -- Scot L. Harris <[email protected]> _______________________________________________ Users maillist - [email protected] http://lists.ximian.com/mailman/listinfo/users