Fritz Borgstedt wrote:
>
>
>>I agree, if all bayesian can do is get the last 1% then it is not
>>worth the
>
>>risk of a false positive.
>
> why that? the last 1% is the most difficult one and it may result in
> hundreds of non detected spams.
>
>
I'm afraid that my "I agree, ..." did not convey what I was thinking.
Baysesian seems to be stopping over 30% of the junk arriving at my mail
server that has the from of legitimate email. The number of false positives
for the last thirty days has been zero. That makes it a useful tool as far
as I am concerned. Drop that to 1% with a high risk of false positive and I
would not thank the same.
There is a problem talking about experience. My data set is small, less
than 10,000 messages, and my test environment is limited to three users in
two domains. 200 emails a day is eighty percent of emails the total emails
that make it through ASSP. It might be that the statistics of such a small
data set can not be compared to a much larger set with any assurance that we
are talking about the same thing.
I am all for less SPAM and just about anyway it can be done. If the current
ASSP 1.2.6 module and configuration I am testing works as well as it appears
to, my users will be happy when I put it on line.
--
View this message in context:
http://www.nabble.com/URIBL-tf2895089.html#a8102212
Sent from the assp-user mailing list archive at Nabble.com.
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Assp-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/assp-user
