From Semantic Web Standards
Jump to: navigation, search

Axel's wishlist on Graph metadata

This is a prioritized/sequential list of how I think the issue of Graph/Triple metadata should be tackled:

1. standardise SPARQL datasets, i.e. be upwards compatibility with SPARQL dataset

  • Propose e.g. TriG syntax or some additional syntactic sugar on top of it, i.e.
GRAPH <n1> { s p o } 

allowed in documents to denote named graphs, and

{ s1 p1 o1 } p2 o .

as N3 compatible syntactic sugar for

 GRAPH _:b { s1 p1 o1 } 
 _:b p2 o2 .

2. extend it by a notion of graph inclusion and semantics for it, e.g.

 g1 rdf:subGraphOf g2 .

(remark timbl in N3 this boils down to g2 log:includes g2 . ).

3. extend the semantics: extend notion of graph merge to dataset merge, extend to RDFS semantics to datasets, what does RDFS mean on the merge of two datasets? bnode scope in dataset merge: note that dataset merge would include talking about bnode scope in named graphs, likely...

4. upwards compatibility with non-named graphs, i.e. a non-named graph as published today is a dataset only consisting of a default graph?

5. other issues (e.g. clarify what "Follow-your-nose" means in terms of this extension)