well it's over 5 years old and even if you're using lots of extra rules I suggest SA will struggle with newer spam around these days.
latest SA is quite different so you may want to start from scratch or at least remove any third party rules as alot of these are now included in the normal ruleset (eg rulesemporium). Also don't forget the install of SA 3.3 doesn;t include any rules at all and you have to run an sa-update to populate with the latest rules. -- Martin Hepworth Oxford, UK On 28 April 2011 14:53, Salvatore <s...@shoponweb.it> wrote: > "John Hardin" wrote: > > You might want to review your bayes training procedures. Do you have > > autolearning enabled? That can cause problems under certain > > circumstances, > > you might want to turn that off. If you are manually training, review > > your > > corpus for misclassified messages. Misclassified messages (e.g. a spam > > that is in your "ham" training mailbox) in your corpus can individually > > retrained after being correctly classified. > > > > You might want to reset your bayes database and retrain from scratch if > > your manual corpus is clean and you have autolearn enabled. > > > > To echo RW, what version of SA is installed? > > I haven't autolearning enabled,(I use MaiaMailguard and I view only > spam-mail for to confirm or not that it's are spam). > My SA version is 3.17, I must upgrade this version for to have best result > about spam catched ? > Thanks. > > > > - > Salvatore. > > > > > >
