This one time, at band camp, David Kempe wrote: >----- Original Message ----- >From: "Peter Hardy" <[EMAIL PROTECTED]> >> I'll also be looking in to generating useful statistics from the mailman >> logs, so we can have a much firmer idea of how effective the current >> strategies are. Is there any packages around that already do this? Any >> advice or suggestions more than welcome. > >I had an idea once that to train these type of packages easily you could put >a hyperlink at the bottom so you could train with a click. Much easier on >admins etc. >Of course it wouldn't make sense for the whole slug list to train. >If you could have a link to a script with specified the message ID which >went to a page where you could chose spam or nospam. Don't know how hard >that all is, but it seemed to me to be a simpler way to train a filter than >the current resending.
The idea Pete came up with involved forwarding the message in question back to a spam-trainer address as an attachment, so that the message can be fed into the spam database of the bayesian filter, and using GPG to ensure that trusted parties do the feeding. I imagine a single script that'd take a message and do the right thing with it, so any list admin could bind a mutt key to do the job, or collect a mailbox in Evolution and feed that in bulk periodically, or similar. -- [EMAIL PROTECTED] http://spacepants.org/jaq.gpg -- SLUG - Sydney Linux User's Group - http://slug.org.au/ More Info: http://lists.slug.org.au/listinfo/slug
