On 2010-04-29 14:58, Raphael Bauduin wrote: >>> The difference is: >>> * BAYES_95 in place of BAYES_05 >>> * score is 6.9 in place of 3.9
http://pastebin.org/192054
As you say the mail has been processed twice, with different configurations or databases, or with the same databases but different users.
Since the headers only contains full scores for one of the passes, it's impossible to know for sure where all the difference in scores came from.
As you noted, one of the passes does have a negative bayes score, while the other have no bayes score, but that's not the only possible difference.
Both passes have AWL scores, but we cannot see what score the AWL applied to one of them. The AWL score may well be quite different between the two passes.
There's nothing strange in getting different total scores when running with different databases and/or configurations. Both bayes and AWL are supposed to be able to give different scores to the same mail in different mail streams.
It is of course possible that there are other differences in scores as well if the two passes were run with different local score settings.
/Jonas -- Jonas Eckerman Fruktträdet & Förbundet Sveriges Dövblinda http://www.fsdb.org/ http://www.frukt.org/ http://whatever.frukt.org/
