W3C home > Mailing lists > Public > public-swbp-wg@w3.org > September 2005

RE: [SE] Suggestion of new note

From: John McClure <jmcclure@hypergrove.com>
Date: Thu, 29 Sep 2005 10:03:22 -0700
To: <public-swbp-wg@w3.org>
Message-ID: <MGEEIEEKKOMOLNHJAHMKCEOKDOAA.jmcclure@hypergrove.com>

A few scattered thoughts....

Perhaps under the title "Worldliness" (!):
1. OWL ontologies define simultaneously an external XML representation for
resources that can be exchanged via text messages, files, or columns in a
database -- OO implies (proprietary) binary protocols for object persistence to
a binary file or an object database.
2. OWL/RDF defines conventions for naming and referencing objects; OO does not
address such issues at all. OWL thus distinguishes between 'first-class' and
'anonymous' resources, while for OO, all objects are anonymous.
3. OWL defines a root instance class -- Thing -- while OO advocates can still be
heard to discuss whether a 'sky-level' class should exist or not (maybe that was
settled, and I just missed it?).
4. OWL schemas are open -- anyone can make statements about another's schema or
any (named) part of that schema.

I think it's important to communicate the THEME that "OWL is Internet-ready OO,
addressing certain issues that OO does not or cannot" rather than the theme that
"OWL is great and OO is deficient" -- a turnoff for sure for certain audiences.
Being "Internet-ready" means providing solutions for issues that simply did not
apply to local and closed binary stovepipes. People like and easily grasp the
notion of being "Internet-ready"... Show historical perspective by tracing the
development of and problems with developing wire protocols for binary objects --
that is, CORBA and the IDL, which was but a standardized OLE (remember those
days?) ...

I suggest a section about the smilarities between OO and OWL, primarily focusing
on the notion that BOTH result in an INFORMATION SCHEMA. So, a section about
"Designing Ontologies vs Designing Object Class Hierarchies" would be amazingly
eye-catching and which could be an opportunity to hark some best practices for
OWL schema development but perhaps more importantly, would be the place to
discuss the degree to which an OO schema can be drawn directly from an OWL
schema and vice-versa.

I also suggest that it be prominently said that OWL (schemas and data) files are
normally processed by an OO-based system, ie fully highlighting the fact that
OWL is not an application language, while OO is, to a degree. This is necessary
I think because the AUDIENCE for this note would be systems architects and
development folks; one area these folks most care about is interfacing OWL files
to their installed base of OO applications and application-building tools, so it
would be on-point to address how an OWL schema can relate to an OO schema. In
other words, it would be good to explain how OO remains, indeed, a key part of
the solution stack for development organizations, with OWL now in the picture.

Finally, I think that a discussion about the impact of OWL on information
querying is quite important, that is, contrast object databases to XML/OWL
databases and contrast the query languages that are being used/developed for
each.

Cheers -
John
Received on Thursday, 29 September 2005 17:01:46 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:31:12 UTC