HI,
from dspam.cgi code,i find that the cgi anaylize $USER.log,and for some
specific signature,it will count total rows including this signature ,and
decide how to show next train action in history page.in this cgi ,variable is
$rclass。
i want to explain this case:
first,some mail first coming into is judged as spam,
after click "retrain checked" or "as innocent" in hitory pages,command
action will looks like as :
/usr/local/dspam/bin/dspam --source=error --class=innocent
--signature=7\,478ef4d4150741031177895 --user
[EMAIL PROTECTED]
then secondlly,i found this message really a spam or i mistake click in
the first step,i retrain the mail as spam again ,command as below
/usr/local/dspam/bin/dspam --source=error --class=spam
--signature=7\,478ef4d4150741031177895 --user
[EMAIL PROTECTED]
all related above is normal command,but my question is that,if in the
second step,i run command on the shell(not from webui) with --class=innocent
,not --innocent=spam as it should be,that is below:
/usr/local/dspam/bin/dspam --source=error --class=innocent
--signature=7\,478ef4d4150741031177895 --user [EMAIL PROTECTED]
how it will happen?in this case ,before run this command,this signature
has treated as innocent ,now i train it as innocent again ,but i set
--source=error!
for me ,i have some specific needs in training dspam,i don't want to
use dspam.cgi,and i don't set quarantine option ,just set as tagged,and i wanna
user can use it in MUA,such as some webmail, and i don't want to analyze
$USER.log file to decide how to show next train action(it will cost too many
resources and not manageable). so, user perhap found this mail tagged as spam
in inbox or in trash according their respective filter settings.
I don't want to decide the next command ,i just wanna provide two
buttons,on click for spam train and on click for inncocent train for the same
mail message,but from log analyze this message may be in any stat(S,I,M,F,count
mod 2 eq 0 or not eq 0),it perhaps be trained as "--source=error
--class=innocent" or "--source=error --class=spam" in the same stat,it can
work? it can keep accuray?
best regards
stevenlee
