On Monday 17 November 2003 01:29 pm, Bart Schaefer wrote: > On Mon, 17 Nov 2003, Brian Godette wrote: > > On Monday 17 November 2003 11:22 am, Bart Schaefer wrote: > > > :0 c : $HOME/.sa$LOCKEXT > > > > > > * ^X-Spam-Checker-Version:.*-$SAVERSION > > > * ^X-Spam-Status: Yes.*autolearn=ham > > > > > > | sa-learn --forget > > > > There's one problem with the above and that's spam that doesn't score > > over the threshold and doesn't hit any rules auto-learning uses. > > That's not a problem with the above; it's a problem with auto-learning > in general. If you can't trust SA's scoring to differentiate ham/spam, > you should turn autolearning off and always train by hand. >
Which is effectively what I do, except I don't have to train a large portion of spam since it is handled automatically leaving only the spam that SA is unsure about and a good chunk of ham to go along with it (lists/personal/ etc). Even with doing this I have a few false negatives every day, but never have had a single false positive on anything other than a bounced joe-job spam which I remove before training. ------------------------------------------------------- This SF. Net email is sponsored by: GoToMyPC GoToMyPC is the fast, easy and secure way to access your computer from any Web browser or wireless device. Click here to Try it Free! https://www.gotomypc.com/tr/OSDN/AW/Q4_2003/t/g22lp?Target=mm/g22lp.tmpl _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
