Bayesian Anti-Spam Filter/Word Classification (Statistical) Class

Bayesian Filter technology (Bayes Filter) is one of the most effective and most promising ways to filter spam (or any unwanted/wanted content). I made this class based on some ideas I found on the internet and some tweaks and tests of mine. It should be very easy to port to other languages and can be used for spam filtering, bulletin boards or chat rooms to flag unwanted (or wanted) posts. It needs to be trained with "good/ham" (meaning wanted) and "bad/spam" (unwanted) posts, and after enough training it can calculate the probability of a text being spam/bad. It also has an auto-learn feature and database support. There are many uses for it, for example, it can be used on PSC to flag unwanted posts, or in RAC to flag posts that are against the guidelines. Or by law enforcement to evaluate chat rooms in search for pedophiles (child molesters), and the like. I've included an access database with a form that can also be used to train the program. I've tested many algorithms in order to improve the code for speed, accuracy, efficiency and reliability. If you have any suggestions or improvements, please let me know. Also, please vote if you like it :)