* Alexey N. Vinogradov <[EMAIL PROTECTED]> [041011 21:22]: > You definitely must not only mark your junk as junk, but also mark your good > mail as "non-junk". Bayesian method works not on the letters, but on the > difference between two sets of them.
Thanks for the mail. DO you mean that even my legitimate mails are caught as spam because I have not taught bayes more ham against spam. (the ratio of my ham:spam training is 2:100) Shantanu -- ________________________________________________________ http://www.silverstones.com/thebat/TBUDLInfo.html