bogus bogotuning

[Peter Bishop]

On 28 Jan 2004 at 17:42, Jason A. Smith wrote:

> No, I just don't see a point in wasting time to do a patch if it will
> never be accepted.  I doubt I would be the only one to use it, I'm sure
> others would try it, even if it produces useless output, it would still
> be educational to experiment.  I'm sure it would produce equally valid
> results with only 1800 instead of 2000 for example.
> 

Could I suggest an alternative?

Why not get some more spam then everybody will be happy!!

Have a look at spamarchive.org
You will find plenty there to get you to the 2000 limit.

Another alternative is to use "bogominitrain"
which leaves the parameters fixed, but is selective about what hams 
and spam are added to to the database maximise detection performance. 

>From the tests it appears to perform well on small numbers of email 
messages and does not need a large corpus.
-- 
Peter Bishop 
pgb at adelard.com
pgb at csr.city.ac.uk