ram wrote: > hi > > what i am looking is > > iam looking sitewide, not userwide > > so if the user feel its spam mail, he will send that mail to another > email of local account, > from there i want to choose the bayes learn and decide what is spam > and what is not spam > > hope i explained well i feel
Yes. Makes much more sense this time! :) You can do something similar to that, but if you do a normal forward, you will generally lose the header information. There are two basic ways to do it. 1) Have the user copy the emails to a local spam folder and then have a process that collects the mail from those folders and learns from it on a regular basis. This is easy to do if you are using IMAP or webmail since everything is on the server. If you are using POP3, it gets more complicated since everyone's mail folder is on their own computer. 2) Have the user forward the mail as an attachment. This will usually preserve the headers depending on the mail client. The downside is that you then have to extract the original mail from the attachment before you can learn from it and you have to teach your users how to forward mail as an attachment. -- Bowie