• KS_ALLOW_REASON "Allow for Multiple Recipients - last Bayesian"

    By Ian Shearer 2 decades ago

    I've been noticing spam's getting past the Bayesian into various mail boxes despite being VERY obviously spam (ie: 'viagra', 'porn', etc in the body). All these messages are tagged with :



    KS_ALLOW_REASON = "Allow for Multiple Recipients - last Bayesian"



    I've searched hi-N-low thru the doc's, this site, etc looking for an explantion on what this means, and how to tune it. It looks like it may be related to the 'split into multiple emails' funciton that was added a while ago, but I'm not sure.



    Tom/ Nico/ anyone - can you shed some light on this



    cheers… Ian

    • Hmmm... No response. Hope KSpam isn't dead - it's a great tool Tom/Nico! <eom>

      By Ian Shearer 2 decades ago
      • By Nico Vis 2 decades ago

        Not at all,

        I am waiting for someone that has some good ideas to add to the project.



        ;-)

    • By Nico Vis 2 decades ago

      Sorry,

      I did not see your question.



      If a spam message has more than one recipient the messages is splitted in one message for each recipient.

      In this way each user can decide a different bahaviour from a single message.

      For example if I have a mailing list for 5 recipients and only one of theese is intrested in receiving it this makes possible to have 4 blocked mail and 1 that passes.

      Unfortunately it programmatically not easy tois difficult to specyfy KS_ALLOW_REASON for each copy of the message so in this cases

      "Allow for Multiple Recipients - last Bayesian"

      means that the received message had more than one recpient and that the last recipient processed had the mail that passed for a bayesian reason (prob<0.9).

      • Thanks Nico, but why no tokens or probability?

        By Ian Shearer 2 decades ago

        Thanks for the response Nico. Great to hear you're still around, and KSpam is alive and well.



        Can I pick your brain a little more? I'm still a little sketchy on what's happening with these messages that are being flagged with "Allow for Multiple Recipients".



        I just got a couple of spam-mails that might be good examples of my question. The 1st spam message that made it to my inbox is not a 'multiple recipients' email. It has been stamped by KSpam with the fields:



        KS_BL_PROB = "0.0000"

        KS_BL_TOKENS="Luncheon: 0.0100"



        This looks fine to me (ie: yes - it should have made it to my mailbox)



        The 2nd email IS a 'multiple recipients' email. There are 4 names listed in the messages 'SendTo' field, and KSpam has stamped this message with only 2 fields:



        KS_ALLOW_REASON= "Allow for Multiple Recipients - last Bayesian"

        KS_IID =



        The body of this message contains "High Quality Name Brand Watches 1500 Styles Of High End Replicas" + a URL. We've seen aproximately 1 trillion messages like this, so I would have thought the bayesian should have trapped on at least "High", "Quality", "Watches", & "Replicas". However there NO probability field (ie: no KS_BL_PROB field) and there are no KSpam tokens (ie: no KS_BL_TOKENS field) on the message. Does this make sense Nico? Shouldn't I see a 'tokens' field and a 'Probability' field? Am I missing something?