For me it works. And actually, I run the sa-learn command only when a SPAM mail gets to my HAM accounts or viceversa (which happens every two or three weeks... maybe more).
If you don't gent false positives nor false negatives, then the Bayesian filter IS WORKING. If ain't break, don't fix it. But it could be that I receive less SPAM than you; I don't know. The thing is, the whole idea of the Bayesian approach is to PERSONALIZE the SPAM detection. So probably some guys (like me) needs to do it every 2 or 3 weeks, and other guys (like you) needs to do it every couple of hours. Canek On Fri, 2004-01-30 at 07:33, Thomas T. Veldhouse wrote: > Canek Pel�ez Vald�s wrote: > > > emerge Mail-SpamAssassin > > > > Read > > > > http://support.ximian.com/cgi-bin/ximian.cfg/php/enduser/std_adp.php?p_sid=bYySnE2h&p_lva=&p_faqid=329&p_created=1039628948&p_sp=cF9ncmlkc29ydD0mcF9yb3dfY250PTImcF9zZWFyY2hfdGV4dD1zcGFtYXNzYXNzaW4mcF9wYWdlPTE*&p_li= > > > > Run every two or three weeks: > > > > sa-learn --showdots --spam --mbox [your SPAM box] > > sa-learn --showdots --ham --mbox [your other boxes] > > > > At the beginning is not perfect; the more you train the Bayesian filter, > > the better it gets. > > > > Canek > > > > Every two or three weeks? I process spam a few times per hour. I have > a script that checks a couple of Maildir folders for the false negatives > or the false positives that I drag into them and then bayes is updated. > My wife sees a lot more spam because she is not as diligent and only > updates her bayes every few days. I suspect that doing this every few > weeks would be nearly completely ineffective [from a bayes approach] > with the way spam changes. -- The only thing necessary for the triumph of evil is for good men to do nothing. - Edmund Burke
signature.asc
Description: This is a digitally signed message part
