Hmm, I had thought that section is generated during the merge process after the GA generates new scores -- it's basically the scores which had been in the previous scores file but which don't appear in the corpus and so were unmodified by the GA (I think that's what it is anyway from memory). The scores were either hand set or evolved by a previous run of the GA against a different corpus. I'm surprised by those 0 scores though; and some of those rules *do* actually appear in the corpus (hunza for example). It's possible that they're falling under some appearance-rate threshold though or something. In general though, I would think it's OK to reset those by hand to something larger based on human intuition not necessarily reflected in the corpus.
C Matthew Cline wrote: > Date: Sat, 2 Mar 2002 17:27:53 -0800 > From: Matthew Cline <[EMAIL PROTECTED]> > To: [EMAIL PROTECTED] > Subject: [SAtalk] 0.0 scored rules > > Looking in the scores files, I find these rules with score 0.0 > > score FREQ_SPAM_PHRASE 0.0 > score FROM_FORGED_HOTMAIL 0.0 > score HUNZA_DIET_BREAD 0.0 > score SEXY_PICS 0.0 > score SPAM_PHRASES_020 0.0 > score SPAM_PHRASES_030 0.0 > score SPAM_PHRASES_100 0.0 > score TO_INVESTORS 0.0 > > They all appear in the "Missed scores" section, which (as far as I can tell) > are scores that weren't touched by the GA. I there any particular reason > they were set to 0, or is that the default score for non-GA'd rules? > > _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
