Hello Alexey, > No. Believe me, the problem exists. since both filters - my BayesIt > and the Bayes Filter use so-called "naive" bayesian method, they are > under the problem of "overtraining". AFAIK the K9 creators use some > solution to fight against it, and it is only the matter of time when > such solution will be realized in the mentioned filters.
I think, the solution is this: you have to keep track, that when a new item was added to the database and limit the database size... This way overtraining can be eliminated. -- Vili ________________________________________________________ Current beta is 3.61.09 (Echo) | 'Using TBBETA' information: http://www.silverstones.com/thebat/TBUDLInfo.html IMPORTANT: To register as a Beta tester, use this link first - http://www.ritlabs.com/en/partners/testers/