Steve Ingraham wrote: > In 2) above you are telling me that 5.0 and even 2.5 is way too high.
Yes. Those are way too high. > So what should it be? It should be zero. This is the default. In other words do *not* set it at all in your files. > Again I do not understand the string of numbers that were displayed > in the Bayes scores that Daryl included in his message. Read the man page for Mail::SpamAssassin::Conf and then put this comment in your config file as a memory aid. # score tag -bayes-net -bayes+net +bayes-net +bayes+net > What about Bayes_95? My big problem is that I am having quite > a few legitimate emails scoring high because of Bayes_50 Most new valid email will score BAYES_50. That is the way that it works. It is either spam or not spam and as far as the classifier knows it is a 50/50 chance of it being either way because it does not have any indication one way or the other. > or Bayes_95 scores. I don't trust a 95% match as close enough to score it too heavily. > What can be done to get these legitimate emails to not be scored on > these Bayes rules? Should I be decreasing these two score > thresholds? If so, what should they be set at? If I should not be > altering these Bayes scores then what else can I do to keep > legitimate emails from hitting on these rules? Train on more valid non-spam emails. Use sa-learn --ham as much as possible to make sure that the statistics are balanced. sa-learn --dump magic The nham and nspam numbers are best when they are about the same number of messages each. Bob
