- From: Smith, Michael K <michael.smith@eds.com>
- Date: Tue, 5 Mar 2002 15:45:45 -0600
- To: Jim Hendler <hendler@cs.umd.edu>, www-webont-wg <www-webont-wg@w3.org>
I wanted to second Jim's comment re rdf:parsetype and N3. I've spent a few
hours today reading documents that are not complete and/or not part of a
standard, after reading email that uses examples with syntax whose
definition I can't find. I sympathize with the desire for compact notation,
but we need to be grounded in those systems we are going to build on.
Towards that end, here's a modest proposal. The goal is to restrict our
options.
I should start by noting that I have been on the wrong side of syntax
arguments in the past. Hard as it is to believe, there are people who
prefer standard infix to God's syntax (otherwise known as s-expressions).
The group I was working with lost a lot of potential interest by only
supporting what to many folks was an unfamiliar syntax.
-----------------------------------------------------------
MOTION: The primary syntax for OWL will be defined in XML.
WHY:
1. We want to lower any potential barriers to adoption of OWL. XML is
vastly ahead of triples or N3 in terms of acceptance by the larger web
community, especially the developer community.
2. Our charter states: "The language will use the XML syntax and datatypes
wherever possible, and will be designed for maximum compatibility with XML
and RDF language conventions." I take the first clause ("will use the XML
syntax wherever possible") as primary.
3. It is clear from reading the DAML+OIL documents that XML was deemed an
appropriate presentation syntax by at least part of this community.
DAML+OIL is supposed to be our starting point.
4. I am assuming that we intend that OWL assertions can be inserted into XML
and XHTML documents. Triples are not going to be the most useful approach.
5. We need to support XML literals. Mixing <ebXML:po>...</ebXML:po> with
triple syntax requires a parser within a parser.
6. I assert (with no evidence) that it is easier to precisely specify the
translation of XML syntax to triples than vice-versa. This could even be
made a formal appendix to the syntax, providing a blessed interface for
existing triple-centric systems.
WHY NOT:
Finally, it has been argued that there are RDF graphs that cannot be
expressed using the XML syntax. I believe that this represents a failing of
RDF to meet the current general goals of the W3C (an historical artifact?).
But this is one potentially serious argument against this motion. The
question would seem to come down to whether those things that we cannot
state in the XML syntax are critical to OWL. Do we have any use cases that
identify such a need? Additionally, this is one of those points noted by
Peter, where what we decide will impact layering.
- Mike
Michael K. Smith
EDS Austin Innovation Centre
98 San Jacinto, #500
Austin, TX 78701
512 404-6683
Received on Tuesday, 5 March 2002 16:50:01 UTC