On Saturday, January 8, 2005, 12:45:50 PM, Kirk wrote:
KM> I've gone through some and haven't found any commonality by sender, etc, KM> but it seems that some are getting through that I'd have expected to get KM> triggered on the subject line alone. For example: KM> Tadalafil Soft Tabs - Great results! KM> ready for the sex life you dreamed about? KM> Buy medications from the net .Pay 1/2 the price of anywhere else KM> Need your prescripiton? We have them KM> VIAGRA $2.99Ljk5 KM> CARTIER, PIAGET, ROLEX Replicas - Expensive Look, Not Expensive Price We do create some filters on subjects, but we try to do that sparingly since they easily cause false positives. A couple of the subjects you've listed above might be in normal conversations on some lists (surprizing as that might be) so we avoid them. Also, many of these subjects change frequently during these campaigns - they coding estimates are that there are more than 50,000 variations on some of the ones you've listed--- and a generalized rule on the subject alone would open up the potential for false positives. That said, I have been coding complex rules for these - the analysis takes a bit longer, but the rules are coming. KM> I've sent numerous examples of each of these in the last week or so yet KM> messages with identical subject lines are still getting through classed as KM> LOW or CLEAN. This just doesn't seem to match the performance that I've KM> become accustomed to, which is why I started questioning whether or not I KM> may have messed up something at my end. Thanks for bringing this to our attention and please do keep pushing us and sending us samples. The more information we have the more quickly we can close in on a set of viable heuristics for these campaigns. _M This E-Mail came from the Message Sniffer mailing list. For information and (un)subscription instructions go to http://www.sortmonster.com/MessageSniffer/Help/Help.html
