John Oliver wrote: > I'm led to believe I can feed untagged spam (and tagged non-spam) to SA > for it to "learn" how to be better. How do I do this?
You were led right. Assuming mbox format mail file -- for instance, Netscape, Eurdora (and my FreeBSD account with qmail is mbox format) - you can feed sa like (with D, for debug output): ./sa-learn -D --spam --mbox mailBoxFile where the 'spam' parameter tells to learn as spam - use 'ham' for non-spam. sa-learn automatically removes SA headers, so tagged or not, does not matter - unless said tags are not SA's. I've read it recommended to take out list mail when ham learning, and I too thought this was a good idea, and have had good results, though the most obvious reason to remove it, is I bypass filtering on list mail anyway. You'll need 200 each of ham, spam, before Bayes will start working - the mail should be from your own account/experience for best results. It's also been recommended that you should have equal numbers of both, and the more the better, but upwards of 5000 (please correct if wrong) is overkill. Me, I'm very light on ham, but still getting very good results. I may change my tune with regards to this imbalance, but so far, no false positives. Though several hundred caught, no spam in my in box today - again. Bryan > -- > John Oliver, CCNA http://www.john-oliver.net/ > Linux/UNIX/network consulting http://www.john-oliver.net/resume/ > * * * * * * * * * * * * * * * > Contribute to the SpamCon Legal Fund!! http://www.spamcon.org/legalfund/ > > ------------------------------------------------------- > This SF.net email is sponsored by OSDN's Audience Survey. > Help shape OSDN's sites and tell us what you think. Take this > five minute survey and you could win a $250 Gift Certificate. > http://www.wrgsurveys.com/2003/osdntech03.php?site=8 -- Nothing in the world has more potential for beauty than woman. Nothing has more potential to destroy it, than the world. - (Anonymous) http://www.wecs.com/content.htm This signature file is generated by Pick-a-Tag ! Written by Jeroen van Vaarsel http://www.google.com/search?hl=en&ie=ISO-8859-1&q=pick-a-tag ------------------------------------------------------- This SF.net email is sponsored by OSDN's Audience Survey. Help shape OSDN's sites and tell us what you think. Take this five minute survey and you could win a $250 Gift Certificate. http://www.wrgsurveys.com/2003/osdntech03.php?site=8 _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
