That sounds good. I think people will need to put in the subject line "SPAM" or "NONSPAM" so that the message ends up in the right corpus though. And even then I bet some of them will end up in the wrong place -- I suppose we could just look at the score/threshold reported for the message and assume that if it's coming to sa-sightings it was incorrectly identified, an so add it to the *other* corpus.
C On 2/28/02 2:15 AM, "Sidney Markowitz" <[EMAIL PROTECTED]> wrote: > "Craig R Hughes" <[EMAIL PROTECTED]> wrote: >> I have to admit that I haven't actually been >> monitoring sa-sightings at all. > > I guess my intuition was right :-) > >> Most of the reason for that is that I don't >> really have the time to manually process all the mails > > Exactly. I don't see how a sightings mailing list is practical unless > its processing is automated. > > I suggest feeding it into a script that knows how to extract MIME > attached forwarded messages. Put the script in the CVS. Then people can > submit patches to teach it to recognize mail that is forwarded inline > with all headers and mail that is bounced, etc., which will allow it to > not waste submissions that are not in the "correct" format, but you > don't have to worry about that at first. > > In any case, forward as attachment is the one way to do it that everyone > can use regardless of mail client, and the one way that is sure to > preserve all the information. > > If your script adds the sightings to a corpus, you can run spamassassin > over it and look over the false negatives. Every time you do that you > will get a list of messages that are either not really spam, so you > should be set up to delete any message from the corpus with a single > click, or are false negatives. > > Would that be automated enough for you to work with? Is it easy enough > to set up? _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
