On Tue, Jul 15, 2003 at 05:08:31PM +0100, Huw Jenkins wrote: > Hi there list, > I'm fairly new to SA so forgive my ignorance. I've installed > SpamAssasssin and it works very nicely. I've got my threshold set pretty > high as we're an ISP and I don't want and False Negatives. I've noticed that > certain pension spam doesn't get in, yet kiddie porn spam falls just under > my threshold (sometime quite far below). Is there a file of words and there > associated scores? I'd like to ramp up words like Lolita etc. > > I may be going about this the completely wrong way! Please tell me if I am > ;-)
Since you're an ISP, people are paying you to provide transit and mail service for them, and to make sure you don't throw away mail they want. It seems to me that you can't afford any false _positives_: non-spam mail that gets tossed or quarantined because you think that it is spam. False negatives just wind up as spam in the user's mailbox; the user can gripe about that, and you can afford a little of it. But false _positives_ will make your subscribers *REALLY* upset. Run the kiddie porn spam through sa-learn --spam, making sure that you also run some ham through sa-learn --ham, and you'll see better results once the Bayesian analyser starts working. -- Mike Andrews [EMAIL PROTECTED] Tired old sysadmin since 1964 ------------------------------------------------------- This SF.Net email sponsored by: Parasoft Error proof Web apps, automate testing & more. Download & eval WebKing and get a free book. www.parasoft.com/bulletproofapps1 _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
