Re: training bayes and newsletters
Bill Cole <[email protected]>
| Newsgroups | gmane.mail.spam.spamassassin.general |
|---|---|
| Message-ID | <[email protected]> |
On 2024-10-15 at 22:05:07 UTC-0400 (Tue, 15 Oct 2024 22:05:07 -0400) Alex <[email protected]> is rumored to have said: > I can imagine the newsletter template is somewhat common, but does > bayes > have any ability to distinguish a junk newsletter from a legitimate > newsletter? Not if it has never seen either of them. Bayesian classification is not magic, but it does do better than random guessing by someone who is not the target of the message, IF it has been trained on similar mail that has been properly classified. > I realize there's somewhat of an imbalance between hams and > spams, but shouldn't there be enough? Absolutely. But the Bayes classifier can't classify mail of types that have been meticulously excluded from its training corpus. > Would I benefit from training known trustworthy newsletters such as > ham? Yes. And train the spam ones as spam. -- Bill Cole [email protected] or [email protected] (AKA @[email protected] and many *@billmail.scconsult.com addresses) Not Currently Available For Hire