On 8/15/2012 5:18 PM, John Hardin wrote:
On Wed, 15 Aug 2012, Kevin A. McGrail wrote:

On 8/15/2012 5:00 PM, John Hardin wrote:

 Right. It might be prudent to review the defaults before the next major
 release.

I wonder if we shouldn't disable auto-learning by default (assuming it's on by default)...

It is.

Bayes should really be trained.

I might not go so far as to say autolearn should be disabled by default, as it is a major good if well trained; but setting the defaults extreme enough that it is reliably, if slowly, initially trained seems to me a fair middle ground. Setting the ham default threshold to -3 or even -5 seems prudent (_much_ better than the current 0.1), then someone who actually wants to configure it can adjust based on how well it's performing and whether they want autolearn on at all.

Can you open a bug about that and let's see if we can get that done? I agree that a slower training threshold makes sense.


--
Kevin A. McGrail
President

Peregrine Computer Consultants Corporation
3927 Old Lee Highway, Suite 102-C
Fairfax, VA 22030-2422

http://www.pccc.com/

703-359-9700 x50 / 800-823-8402 (Toll-Free)
703-359-8451 (fax)
kmcgr...@pccc.com

Reply via email to