Section 5: LDPC formal definition and limitations

Section 5 defines LDPCs without specifying formally how they relate to
RDF resources and LDPRs. Is the following true?

  ldp:Resource rdfs:subClassOf rdfs:Resource .
  ldp:Container rdfs:subClassOf ldp:Resource .

If not, what explicit relationship is there? Also, what is the
RDFS/OWL definition of a "membership triple"?

It is also pretty straightforward to map LDPC membership to SPARQL
triple patterns:

  ?container rdfs:member ?member
  ?membershipSubject ?membershipPredicate ?member

You can see how limited they are. In the first case, container is
required to maintain explicit membership triples, which is expensive
and not present in most of the existing RDF datasets.
In the second case, explicit membership properties do not allow
containers where the subject is unbound, for example, where members
are instances of specific type

  ?member rdf:type o:NetWorth .

By restricting the containers with fixed subject and predicate, the
spec artificially creates a need for per-predicate containers, which
adds a unnecessary complexity in the form of base/relative URIs,
composition/aggregation etc.

I suggest that the membership section is redesign based on triple
templates and/or SPARQL queries.

Moreover, what are the formal RDFS or OWL descriptions of
ldp:membershipSubject and ldp:membershipPredicate properties? How do
they relate to existing RDF/RDFS/OWL concepts? What are the domains,
ranges, and super-properties?

Martynas
graphity.org

Received on Friday, 22 March 2013 10:37:35 UTC