neural networks and filePro

[John Esak]

> The examples I've seen so far are nothing more than raw correlative
> matches, not anything based true personality analysis.  Hell, people can't
> even accurately judge others' personalities--I surely wouldn't
> trust such a
> subjective job to a computer, which will be easily deceived, and which has
> no concept of motivations.

You know... not to disagree... but, if you are actually on the AOL site, and
you purchase a CD for the very first time of any group of purchases... in
other words a new purchase, new session, etc., and they have no previous
data on file for you... the "suggestions" they make as to what you might
also like are surprisingly and quite amazingly good. Meaning, they are not
just in tune with the most popular stuff, etc., but you can buy sometthing
pretty eclectic, say and old Bob Dylan CD and the suggesions are things you
might _actually_ like. This alone, tells me that the people grouping the
genres and actual artists are doing a pretty good job.  At least, that has
been my experience... many times.  As to how much my previously purchased
material adds in to their next suggestions, I'm not sure... haven't really
seen this strike any big chords yet.

John