|
|
|
|
|
 |
Tuesday, September 17, 2002 |
|
I've been slashdotted. Thanks! :) 10:26:02 PM 
|
|
Paul Graham's essay on a Bayesian spam filter has been getting a lot of attention and is the basis for several software projects. However, even though it works well, I believe it could work even better.
The problem interested me, not least because I seem to get 5 or 10 spams an hour. So I have written up some potential improvements to the Graham that are untested at this point, but are justified by theory. If anyone with spam detection code wants to test them and get back to me, I'd appreciate it. |
 |
 |
|
 |
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||