2007/2/12, Matti Haack <[EMAIL PROTECTED]>:
Hello,
can the baysian confidence be change so, that in Baysian testmode
"Prepend Spam Subject" is only added if message is spam and confidence
is higher than Bayesian Confidence Threshold?
Then of course the mail should not be put in the spam corpus if the
confidence is lower than the threshold.
You wouldn't know when a correction is necessary because the subject isn't
tagged.
The problem is that mails with low confidence are the most interesting for
fine-tuning the bayesian analysis because they are new kinds of spam and
ham. You do of course still get new kinds of spam added in your corpus by
the other checks but no longer by the bayesian check that needs human
intervention to correct mistakes.
Probably this would lead to:
- The Bayesian filter passes more spam but tags (almost) no false positives.
- The Bayesian filter learns less and slower but learns (almost) fully
automatic.
I use the baysian filter in tesmode only, because I won't block any
mail to my users but like to prepend it with a spam tag. This works
fine.
Now I experimented with baysian confidence, as I have some false
positves with a confidence 0.0000 - Which I want to be excluded from
tagging.
and excluded from adding to the spam corpus of course...
-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier.
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
Assp-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/assp-user
