> James currently touts a Bayesian mailet, but employs only an > overall data source and is not concerned with individual > preference; to be an effective SPAM blocker, a relationship > needs to be established between a specific user and her > Bayesian lists.
Actually, Alan, you are mistaken; James does not tout Bayesian Analysis anywhere on the website or in any publicity or documentation. There is code around, shared with this list by Chris Means, which quite effectively performs bayesian analysis of mail, but it is not yet part of James, and won't be until it is easy to deploy, configure and use. The code in question is a mailet, and can be configured per-instance to use a particular repository for its corpus. This allows its use on a per-account or per-address basis, deploying it in this way would be up to the administrator who configured the James instance. > and of course the necessary functionality to maintain the > lists ... Chris Means code includes this, and training is very effective, if a little time-consuming. I hope to be able to simplify the administrative process, and add the ability to use filesystem storage as well as database for the corpus and tokens. It is my hope that Chris' Bayesian Analysis mailet will soon be added to the James v3 code under development. d. --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
