Relationships between terms, statements and schemas and proper application behavior

I just wanted to toss the following comment out, which I didn't
get an opportunity to do during the f2f when discussing the
relationships between terms, statements, vocabularies and schemas.

It was suggested that all that was needed was the relationship
between term and schema, and that if an application needed to
know what statement was said about the term by a given schema,
then it could go look at the RDF/XML of that schema itself.

I consider that to be an unacceptable requirement on the
application and if done, to be unacceptable behavior by
the application.

If there is knowledge present in an RDF/XML schema that is
intended to be visible to some RDF application, such as the
provenance of a given statement per a particular schema,
and that knowledge is not carried over into the graph, then
that is a bug (of some sort).

If it's not in the graph, it doesn't exist. RDF applications
should *not* go back to the RDF/XML serializations to try
to grok knowledge that is not explicit in the graph.

The parsing of RDF/XML into triples defines *all* of the
knowledge that a given RDF application is licensed to
consider, and if a conformant parser has not generated a
given triple itself, then the application should *not*
infer that triple from the RDF/XML serialization separately.

Thus, if specific statements must be associated with specific
schemas where they are expressed, then the mechanisms provided
by RDF to do so must be used -- namely reification.

If each statement in a schema is not explicitly reified with an
rdfs:isDefinedBy qualification pointing to that schema, and
all that is provided are rdfs:isDefinedBy qualifications for
terms, then *all* that can be known by a given application is
that the schema says *something* about the term, but it can
never know exactly *what* is said about the term.

Cheers,

Patrick

--
               
Patrick Stickler              Phone: +358 50 483 9453
Senior Research Scientist     Fax:   +358 7180 35409
Nokia Research Center         Email: patrick.stickler@nokia.com

Received on Thursday, 20 June 2002 03:32:55 UTC