Re: KR and RDF (was RE: subclasses (RDF vocabulary definitions))

1. Primary design goals of KR
    simple yet powerful -- can describe anything
    English-like, but unambiguous
    support genus-differentia definitions, concept-hierarchies, context

2. Making a very gross analogy, KR is like UNIX -- common words are short (my choice) & somewhat cryptic (not too bad I hope),
and RDFS is like Windows -- common words are long and easy to understand.
As I mentioned previously, KR will accept RDFS long words, if desired.

3. The reason that I said OWL should be the starting point
is because it has a sufficient set of concepts to fairly easily describe anything
============ 
Dick McCullough 
knowledge := man do identify od existent done
knowledge haspart list of proposition

  ----- Original Message ----- 
  From: Jon Hanna 
  To: www-rdf-interest@w3.org 
  Sent: Friday, November 22, 2002 6:46 AM
  Subject: KR and RDF (was RE: subclasses (RDF vocabulary definitions))



  > 4. If you compare KR and RDFS, I think you will find that
  > a KR description is 10 times shorter and 10 times easier
  > to read.

  Easier to read for whom?

  I think this is were KR and RDF will differ, and probably should differ.

  If I understand correctly (I haven't had time to play with KR yet) KR is
  designed explicitly to be reasonably close to English, and hence
  understandable to human readers. RDF is designed to define relationships
  between resources (referred to by URIs, though applications are as likely to
  use 32-bit numbers internally) that is easy for machines to use.

  While human-readability has been a feature of the design of URIs, and of the
  various ways we express RDF (n-triples, RDF/XML, graphs in the sense of
  actual pictures) it is of secondary import to RDF.

  Neither of these makes one better than the other, but it does make one
  better than the other for certain tasks.

  > P.S. One of the things that makes RDFS more verbose is the
  > multiple layers of definition, which then require qualifiers
  > to distinguish the names, e.g.: rdfs:Class instead of Class.
  > The OWL level should be the starting point, and should be the
  > default with no qualifiers required for its names.  That's my
  > practical advice for language design.

  That's verbosity of particular forms of expression, not of RDF itself. In
  RDF rdfs:Class is a shorthand for
  http://www.w3.org/2000/01/rdf-schema#Class. In turn that URI identifies a
  resource (the concept of Class as defined in RDFS). This being the web URIs
  identify absolutely everything. Remember RDF isn't a language per se. It is
  a framework for statements concerning triples of web resources that we use
  different languages (RDF/XML, n3, diagrams, and a bastardised pseudo-n3
  common in discussions on this list and elsewhere) to express.

  The rdfs: in rdfs:Class is just a convention. Where appropriate you can use
  Class, s:Class or nowImJustBeingSilly:Class. Whether OWL, RDF, RDFS or a
  completely differnt level is the most natural "starting point" varies from
  case to case.

Received on Friday, 22 November 2002 11:23:41 UTC