Carlo Wood wrote: >On Wed, May 04, 2005 at 12:03:15PM -0400, Matt Kettler wrote: > > >>If you're using qmail-scanner's fast_spamassassin option, that's your >>problem. >> >> > >Thank you! I had a qmail-scanner 1.21 which didn't have >a verbose_spamassassin option. I got 1.25 and configured it >with --scanners=verbose_spamassassin and now things work as >expected. > > >A problem with the filtering seems to be Bayesian classifier, >I get: > >Spammy tokens: 0.970-+--H*Ad:D*alinoe.com, 0.965-+--HTo:D*alinoe.com, >0.960-+--H*Ad:U*carlo, 0.957-+--H*Ad:D*com, 0.953-+--HTo:U*carlo, >0.949-+--HTo:D*com, 0.899-+--st0ck, 0.891-+--H*RU:alinoe.com > >In -well- every mail. That is not too weird, since >this is my domain! Why does rate 'alinoe.com' and 'com' >and 'carlo' as spammy tokens? Is that normal? > > > No, it's not normal.
Have you been training your bayes using forwarded messages? In general it looks like your bayes has been very heavily trained on spam that was addressed To: you, and almost no nonspam messages addressed To: you. This is something that could happen if you were forwarding mail for training, or if you used someone elses nonspam for training (and little or none of your own), but did use your own spam.
