W3C home > Mailing lists > Public > public-owl-dev@w3.org > January to March 2007

Ontologies and ECMA data models

From: John McClure <jmcclure@hypergrove.com>
Date: Tue, 23 Jan 2007 13:08:06 -0800
To: <public-owl-dev@w3.org>
Message-ID: <MGEEIEEKKOMOLNHJAHMKCENPEDAA.jmcclure@hypergrove.com>

I'd like your opinions about leveraging ontologies in a more useful way.  I
do not propose changing the way that ontologies are defined, but rather how
RDF triples are created from instance declarations.  Assume two classes, X
and Y.

<owl:Class rdf:ID='X'>
  <owl:ObjectProperty rdf:ID='x'>
     <rdfs:range rdf:resource='Y'/>
<owl:Class rdf:ID='Y'>
  <owl:DatatypeProperty rdf:ID='y'/>

RDF specs envision XML coding as:
  <X rdf:ID='R01'>
     <x><Y y='string'/></x>

ECMA language coding would simply be:
  myX = new X();
  myX.x.y = 'string';

The underlying processor can know by reference to an ontology that y is a
property for resources of class Y which is the range of the x property for
instances of the X class. From this simple string, the XML coding above (or
an R3 coding) can be generated without any trouble whatsoever.

Ontologies today seem useful only to define a vocabulary whose terms are
valid during exchange of XML/R3 files while the W3's primary focus seems on
enabling reasoning-capable software which identify more precisely the set of
classes applicable (and, for validation, not applicable) to an instance.
But if an IT organization is not now interested in reasoning software, and
it finds little immediate economic justification for coding per an exchange
standard, then why in the world would the organization care to create or use
an ontology?

So, ECMA languages -- Jscript and C# -- can offer a compelling OWL adoption
story for these organizations *IFF* ontologies are positioned as the means
by which ECMA object models are defined and processed.  The RDF's reasoning
capabilities are then better positioned as another rung up the maturity
ladder for the IT organization to be adopted according to the organization's
own internal criteria.

For this group one impact of this re-positioning is that the RDF/A
specification should allow an ECMA-oriented representation of the name of
the text property that a string in the X/HTML file represents. Coding like
<span property='Author.Family.Name.eng'>McClure</span> should not only be
recommended by the SWBP, but should receive the strongest organizational
endorsement possible from the W3 because -- from all apparent evidence --
this appears to be the only way that the RDF will be adopted on a wide-scale
global basis.

Let's examine <span property='Author.Family.Name.eng'>McClure</span>.
Historically each of the three phonemes (Author, Family, and Name) have been
construed by ontologists to be names of classes.  I disagree -- it is much
more powerful if these are the names of object properties, while lower-case
names (eng) are datatype properties. IOW, ontologies could be better
engineered to facilitate ECMA-based programming tasks; a class and property
naming recommendation is absolutely needed; and a primary organizational
focus on RDF/A is critical to the success of the Framework as a
widely-adopted technology.

Thanks for your comments.
Received on Tuesday, 23 January 2007 21:07:03 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:58:14 UTC