> -----Original Message----- > From: Fuzzy Fox
> The only thing that worries me when I hear people say "Just > let Bayes do it" is that it makes me think this is rolled into the > singular Bayes score check that SA already does. > > It seems to me that this would be more effective as a separate, > Bayes-trained database, which simply keeps track of spam scores > received at various times of the day, and tries to determine, based > on what time it is, what is the probability of spam? > > This would become a separate score (positive or negative) in > SA's normal scoring process. And Bayesian word-tokens would be a > different, separate score as well. I wish I knew more about the guts of the Bayes implementation and training to better comment on this. However, I would tend to agree with you on this. This is entirely a different test and would be best separated from word-tokens. This would better keep the tests compartmentalized and reduce the collateral damage from updates. --Larry ------------------------------------------------------- This SF.Net email sponsored by: Free pre-built ASP.NET sites including Data Reports, E-commerce, Portals, and Forums are available now. Download today and enter to win an XBOX or Visual Studio .NET. http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01 _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
