Re: LYNX-DEV Internal MIME types

[Al Gilman]

  From: "Christopher R. Maden" <address@hidden>

  [Al Gilman]
  > This is a positive step.  I am not sure ...

[Chris Maden]  
  Ah - but there you're wrong. (-:  

Whoa! if I'm not sure, how am I _wrong_?  Particularly since...

  This is something that's being put off for a little while, 
[snip]
  Any suggestions anyone has are more than welcome.  This is really an
  issue with generic information classification, and one that has never
  been adequately solved.  

Yes.

People keep confusing data with information.  Information is the
capability to satisfy queries.  If you work on a projective way
of subdividing the space of queries, you get somewhere.

                           Dewey Decimal, Library of Congress, Dublin
  Core, Roget's, and PICS are all relevant buzzwords here, but what
  combination, if any, will provide the answer is unknown.
  
No combination of the above.  Two more fruitful avenues of approach are

        Coputational:  Consider what makes the Excite relevance filter
                a winner, and proceed on from there.

        Formal:  The class description methodology has to blend rules
                and lists of instances gracefully.  Both intension
                and extension.  A recipe is roughly the following:
                Take ARPA KIF, extend from lambda-abstraction to
                kappa-abstraction, blend with CORBA objects and go.

$20,000.00
--
Al Gilman
;
; To UNSUBSCRIBE:  Send a mail message to address@hidden
;                  with "unsubscribe lynx-dev" (without the
;                  quotation marks) on a line by itself.
;