Luis Hernán Otegui wrote: > First, RTFM. > Second, Google. > Third, oh, well... You NEED to feed Bayes a significant amount of > data, so it knows what is spam and waht is ham, due to the fact that > the kind of spam and ham you receive is different from the ones I get > on my servers. Then it will start auto learning on that basis. But, to > start, it needs you to feed it data... > > Luix > > 2007/5/11, Daniel Aquino <[EMAIL PROTECTED]>: > > > Have you trained the bayes database? Is this a fresh install? It > > > needs at least 200 spam and 200 ham messages to get it going. > > > However, the more ham and spam you can feed it, the better it > > > will perform... > > > > Well I thought I could use the auto-learning feature ?
You can use auto-learning, you just have to watch it at first and manually re-learn any messages that it mis-classifies. Also, depending on your traffic patterns, it can take a while for Bayes to learn the required 200 spam and 200 ham messages from auto-learning alone. For best results teach it manually at the beginning. This way, you know you have a good database when you start. With auto-learning alone, the database can get corrupted before you ever get a chance to use it. -- Bowie
