A week ago I attended the W3C Technical Plenary Week in Cannes
(France). There was an very interesting session on Formal Methods at
the W3C [1]. The specs that had benefited from working with formal
methods were:
- XQuery
- OWL and RDf
- Web Services Choreography and Web Services Descriptions
(using pi notation)
- WSDL 2.0 (using Z notation)
In some cases parts of the formalization became the authoritative
work, in others they only had a guiding and clarifying role, and the
english text remained the authoritative work. But in every case the
feeling was that the formal backing helped work out a lot of problems
that were difficult to notice in the natural language text.
Now each of the formal methods languages used is in the end back by
Set Theory, which RDF and OWL and other semantic web enterprises are
based on. My suggestion was therefore that RDF/OWL could itself be a
formal language one can use to help explicate various standards.
The work being done by the AtomOwl group could then be understood to
simply be a formalization of the work done by the atom syntax and APP
groups, where the spec release by these groups remain the
authoritative texts. The advantage of the Semantic Web languages over
these other formalisms is that it is much more widely understood, is
web centric, and furthermore has some very interesting direct
practical applications. AtomOwl [2] could for example be directly
used to help create Sparql end points for aggregators [3], as well as
being used directly as a data model as BlogEd does [4].
It is in this light that I am looking to set W3C Working Group (or
XG) to create a standard ontology for Atom. As a formalization of
Atom it will have to be faithful to the specs released by these IETF
groups, which as stated, remain authoritative. There is interest from
IBM, Talis Information and Sun Microsystems, three members of the
W3C to work on this. If other members would like to join we would be
more than happy. We would be very interested to hear from other
members of the Atom group on how we can proceed to make the resulting
spec be of the highest possible quality. Ideally we would like to
place the resulting document at the Atom namespace location with some
(XSLT?) transform that would be the default transform from atom to RDF.
We are very open on what the best way to proceed is.
Yours sincerely,
Henry Story
Sem Web Researcher, Sun Microsystems
http://blogs.sun.com/bblfish/
[1] Session 7: Adventures in Formal Methods
http://www.w3.org/2006/03/01-TechPlenAgenda.html
[2] http://bblfish.net/work/atom-owl/2005-10-23/
[3] see http://www.intertwingly.net/wiki/pie/PaceSparqlLink
[4] https://bloged.dev.java.net/