With reference to the Bayesian filter, I am referring to 'pattern matching' as the collection of data used for it's analysis.
This boils down to a collection of 'good mail' and a collection of 'bad mail', that in my opinion needs to reflect the users' interests, therefore I am a bit leery in 'sharing' this data, but am not dismissing it's potential. The advantages of this system is that the filtering process will grow with the user, but what's key, is that the user has control over the filtering process. By expanding on Chris Means' mailet to allow for user specific corpus' and then adding the functionality of receiving forwarded mail to the two lists, we end up with a simple yet robust system that adds value to the end users .. > can you expand upon what you consider "pattern data"? > > b > > alan.gerhard wrote: > > back up a bit - > > > > my point differs in that the pattern data collected is > > individual and i do not see too much need for sharing. > > other than that, the outstanding issue is, as a james > > user, how to go about setting up and maintaining > > Repository based data for the Bayesian mailet. > > > > not to cut this discussion off - i just want to > > highlight an addressable issue and explore different > > solutions --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
