• Bayesian: I have >7000 mails in Mailspam, is this advisable

    By Jörg Asmussen 2 decades ago

    Is it good to have many mails in mailspam.nsf?

    Should I purge old mails?

    How many mails can be in mailspam before fidelity/ performance ratio decreases?

    • Number of messages in dbs

      By Nico Vis 2 decades ago

      We have spam containers of more than 15000 docs. more than 500 messages are blocked each day. We manage it with space savers in the replication settings tab. But actually it is a good question.

      Tom, which should be the right amount off messages for mailspam and mailgood?

      Thank you

      • 50/50

        By Markus Russ 2 decades ago

        Hi



        somewhere in this form i found a 50/50 ratio, for myself i kill all mailgood >14 days

        and reduce the size of the mailspam by agent older than 10 days.



        the db size is about 500 mb each and include about 6000 docs.



        avoid the mailspam getting much bigger than the mailgood, the result is more false positives.



        don´t you guys having problems with false positves ?

        as i post 3 posts deeper, lots of html-web mails are stamped as spam, the solution

        with the notes.ini param ks_bl_ignore is not a good solution, cause there are lots of

        html code to exclude ….



        do you guys check the spam for false positives ?



        we have about 600 spams found by kspam - 20-30 are false positve ……

        … mostly customer orders, without maintenence by hand this solution

        is not possible.

      • Are messages in MAILGOOD being used at all?

        By Jörg Asmussen 2 decades ago

        I thought mailgood was only a convinient place for manual checking of false negatives.