Re: Taking an axe RDF in XML? (no thank you)

>>>>> "AP" == "Assini, Pasqualino" <of Thu, 23 May 2002 15:43:48 BST> writes:

    AP> I am not sure that the XML syntax is a major stumbling block on the
    AP> road to the adoption of RDF.

    AP> ...

    AP> My explanation of XSLT success vs RDF failure to gain widespread
    AP> acceptance is that the use case that XSLT is made to provide,
    AP> tranforming XML, is both widely needed and easily understood. 

    AP> ...

    AP> On the contrary: you describe some resource in RDF and ... not much
    AP> happens. 

    AP> ...

    AP> But I would certainly agree that an aesthetically pleasing syntax that
    AP> people could write/read easily would help enormously in
    AP> kick-starting the semantic web.

Both points are valid: use cases and syntax.

1. Use cases.
   There are no convincing applications of RDF per se.
   This is just a language for specifying facts. No reasoning (worth of
   this term) is possible in RDF as such.
   To make it more attractive, it should be part of a larger specification
   (e.g., rule-based) where interesting things are possible.

2. Syntax.
   There are two issues here: the XML syntax and the "surface syntax",
   i.e., the syntax in which  people will actually write specifications.

   a. XML Syntax: In my opinion, this effort was wrong-headed from the start.
      XML is just a serialization syntax for the machines to handle.
      It is not human friendly and doesn't need to be. People should write
      specs in a surface language suitable for people, not machines.
      The translation should be done by the tools.
      If one agrees with this point of view, then the XML syntax should be
      as close to the surface syntax as possible -- just add tags to make
      it easy for the machine to parse. There should be exactly one
      translation from the surface syntax to the XML syntax -- not half a
      dozen as it is today.

   b. Surface syntax: This (rather than the XML syntax) should have been
      part of the original specification. I would like to point to XQuery
      as an example. They first set out to figure out what the language
      should be, what is the semantics, what are the use cases.
      The XML syntax was left for desert. (To be sure, some in the XQuery
      community tried to put the carriage in front of the horse, too, but
      fortunately a more sensible approach prevailed.)

      What should be the surface syntax for RDF? N-triples is ok and is
      easily serializable), but there are better and more established ones.
      Stefan Decker and Co. has been advocating F-logic syntax for quite
      some time, which is clearly superior to N3. It is part of a larger
      theory (wich can give you convincing use cases--see 1), has natural
      semantics, and is easy to write and understand.
      


Michael Kifer
kifer@cs.stonybrook.edu

Received on Thursday, 23 May 2002 11:31:29 UTC