James G. Sack (jim) wrote:
<heh> I have the same reaction to "folder".
I used to but I'm afraid years of Windows-speak has worn me down. When
I'm in a shell I still refer to them as directories. When I'm in a GUI
and the thing is represented by a folder icon and I'm talking to a
non-technical user I refer to them by folders also.
.. But, I think the /virtual folder/ part does have a conceptual
strength in allowing your view of the information _organization_ to be
Shouldn't it be called a "virtual directory"?
For email, it strikes me that there is one unquestionably superior
indexing method (by receipt timestamp). Sender and subject (etc) are
great secondary indices, but I'd kinda like to be able to apply multiple
and time-varying ad-hoc tags to email messages. Maybe even have some
kind of bayesian autotagging?
Funny you should mention bayesian autotagging! On my list of things to
learn for years now has been CRM114 (google it). It is a sort of
programming language with bayesian functionality built in. You can use
it to classify text. You could use it to automatically sort mail.
Manually sort for a while, let crm114 analyze how you sorted it, and
then from then on it will find the best place for incoming mail. It is
also potentially useful for doing bayesian filtering of system logfiles
sending you only the non-routine stuff. This has all sorts of potential
uses. This is finally useful artificial intelligence.
I'm even more ignorant about bayesian filtering, but I'd like to know
more. Maybe Andy will give us a lecture on it one of these days?
There is a good article on it on wikipedia:
http://en.wikipedia.org/wiki/Bayes_theorem
A bit technical but it boils down to a basic mathematical formula which
allows you to calculate a probability based on new evidence without
having to recalculate all of the old statistics. It's like keeping a
running tally.
--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
