-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 The Tuesday 2006-08-01 at 23:54 +1000, Basil Chupin wrote:
The question is why is the spam scanner finding so many "hits" in some messages?
The Bayesian filter is badly trained, perhaps as a result of the automatic training feature it has, or because it is not used for a single destination but company wide. This hits some suse lists badly, like the Spanish language list, suse-linux-s.
And if it is tagging messages as SPAM then why is it allowing them through to be posted in this forum? Why have a filter which does not filter :-) ?
I think this initial filtering is company wide; it's up to each destination (user, employee, etc) to act on it or not - that is the correct behaviour, by the way (IMO). As I understand, if my memory serves me right (I corresponded with suse people about this problem at the time), many people got their email to suse lists rejected because they were wrongly marked as spam, and could not post. As a "temporary" solution, let's see what happens kind, they decided to let it mark spam, but not remove them. All this will be modified when these lists get moved to a new system, the one that is used for opensuse lists. - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2 (GNU/Linux) Comment: Made with pgp4pine 1.76 iD8DBQFE5c1ntTMYHG2NR9URAkpbAJwJni8rb/zO812NH8jLrbE2JOZ3LwCfe5kU vhd19rY6InsE7kGank5Km98= =OxOW -----END PGP SIGNATURE-----