Hello Alexander,
Sunday, August 8, 2004, 4:36:01 AM, you wrote:
ASK> Hello Peter Kerekes,
ASK> 08-Aug-2004 00:21, you wrote:
>> Need some help to improve on Spam filtering.
ASK> Forgive me the stupid question, but do you actually "train" the BayesIt
ASK> filter if it has a false negative? (right click on the mail, Special, Mark
ASK> as Junk).
Sure. First I did train on approx. 600 junk mail, all addressed to me, and
keep on training since then continuously.
>> Can anyone suggest a change in any settings to improve filtration?
ASK> Have you trained the BayesIt filter by importing a large Spam database? I
ASK> found its accuracy dropped when I did that. I only used "my" Spam for
ASK> training BayesIt, and its accuracy was much higher.
--
Best regards,
Peter Kerekes (Toronto, Canada)
"When two people in a business always agree, one of them is unnecessary."
-- Anon
TB! v2.12.03 and BayesIt! 0.5.9 on Windows 20005.0.2195 Service Pack 4
________________________________________________
Current version is 2.12.00 | 'Using TBUDL' information:
http://www.silverstones.com/thebat/TBUDLInfo.html
