What I Want in RDF 2.0

Position Paper for W3C RDF Next Steps Workshop, June 26-27, 2010, Stanford, CA, USA
David Booth, Ph.D.
Cleveland Clinic (contractor)
Latest version of this document: http://dbooth.org/2010/rdf2/

1. Standardize a rules language based on SPARQL CONSTRUCT.  (For example, see SPIN.)  Even if other rules languages are ultimately standardized as well, a SPARQL-based rules language is a no brainer, since RDF users already know SPARQL.  If such a rules language were based strictly on SPARQL CONSTRUCT -- nothing more and nothing less -- the standardization effort would be minimal.

2. Standardize an XML Schema-friendly serialization.  And of course it must support named graphs.  TriX is the most obvious candidate.  Because RDF/XML is the only standards-based serialization for RDF, people feel compelled to use it, in spite of how awful it is.

3. Permit literals as subjects.  Although there are work-arounds, they add unnecessary complication, confusion and debate.

4. Named graphs.  Being standardized will help push tools to uniformly support them.

5. Do something about bnodes!  (Somehow.)  Get rid of them?  Maybe provide a standard namespace for minting recognizable, but unique URIs?  In other words, a standard way to skolomize them.  E.g., http://w3.org/bnode/dbooth.org/2010/06/26/fribjam432

6. Standardize turtle, n3 or other human-friendly syntax.  Reading RDF/XML is like trying to read hexadecimal.

What helps us (experts) is far less important than what will help new users!