antispam system on tarsier reloaded
I have finally got some progress of the Bayesian traning on tarsier. Now it has learnt the data from the past six months, both wrongly detected hams and spams. I hope this can improve the accurancy of the anti-spam system.