[Spambayes] New Application of SpamBayesian tech?
skip at pobox.com
Fri Feb 28 20:08:14 EST 2003
Todd> I'm the person you quoted in your posting so I'm very intrigued by
Todd> SpamBayesian tech? Can you explain to me what this is in layman's
Essentially, given a pile of "appropriate" and "not appropriate" documents
(spam vs. non-spam email messages so far, but it could be resumes or
appropriate vs. not appropriate web pages), the system is trained using
them. The system tokenizes the document (not always in a completely
straightforward fashion) and counts how often various tokens occur in the
spam vs. non-spam documents. Training complete, unknown documents are fed
to the system and it classifies them based upon the relative "spamminess" or
"non-spamminess" of tokens in the document.
There's a much better explanation on the SpamBayes website:
More information about the Spambayes