Re: Common RDF parser bug?

Jeremy Carroll wrote:

> > This does mean that anyone using an RDF parser to load an RDF-XML
> > document has no guarantee that the RDF is valid. It will certainly seem
> > counterintuitive to RDF newcomers that a conformant RDF parser will
> > allow invalid RDF.
> >
> > It also seems from the replies that different approaches to such invalid
> > RDF will occur; some people may attempt to infer equivalences, as Sandro
> > suggests, but others could equally well reject the RDF as invalid after
> > detecting the wrong number of properties. Which approach is _correct_?
> > If there is more than one 'correct' approach, then the meaning of that
> > piece of RDF is ambiguous. I think this needs clarifying.
> >
> 
> On the basis of the new work coming out of the WG I would assert the
> following:
> 
> Given an RDFS schema, we can take an arbitrary (i.e. schema invalid) RDF
> graph and find a smallest supergraph that is RDFS valid. This is its schema
> closure.
> 
> Thus every RDF-valid document is RDFS-valid.

I'm not sure what you mean by RDFS-valid. I am talking about graphs
which are not even valid RDF, because they have more than one
rdf:subject, rdf:predicate, or rdf:object property attached to a node
which is an rdf:Statement. Just according to the RDF M&S such RDF is
simply not valid, as far as I can see.

> Cardinality constraints cannot be expressed in RDFS. Domain constraints end
> up implying that the subject of rdf:subject is a Statement.
> 
> The cardinality constraint in question requires a higher level ontology
> language (such as DAML+OIL) to express it. I do not think an RDF (or RDFS)
> processor should attempt to validate such constraints. 

I understand your rationale, but do you see my point that a user of an
RDF parser would reasonably expect it to produce a valid RDF graph as
defined by the M&S in all cases?  Are these constraints not an integral
part of the RDF standard? If such checking is not in the parser, where
should it be?  Are we saying that to make any use of RDF, one must
always include a DAML-like reasoning engine as well as an RDF parser? 

The constraints here are quite specific, so I am not clear why they
should be pushed out to a higher-level language rather than just being
implemented directly in code.

> Hence, a system that
> offers such validation, should IMO, make it optional and clear to the user
> that this is an additional validation over-and-above the standard RDF
> validation.

Well, what counts as "standard" RDF validation? 8-)

Regards,

David Allsopp.

-- 
/d{def}def/u{dup}d[0 -185 u 0 300 u]concat/q 5e-3 d/m{mul}d/z{A u m B u
m}d/r{rlineto}d/X -2 q 1{d/Y -2 q 2{d/A 0 d/B 0 d 64 -1 1{/f exch d/B
A/A z sub X add d B 2 m m Y add d z add 4 gt{exit}if/f 64 d}for f 64 div
setgray X Y moveto 0 q neg u 0 0 q u 0 r r r r fill/Y}for/X}for showpage

Received on Monday, 21 January 2002 08:40:06 UTC