Re: a blank node issue

Hi Michael,

Good idea putting this together :)

Michael Schneider wrote:
> Within the last few days my mailbox must have been hit by more mails about blank node issues as there are different blank node identifiers on the Web. It's these old discussions again and again: Should the syntactic scope of blank nodes be graph-local or global? Should the semantics of blank nodes be that of existentially quantified variables or skolem constants? Should there be blank node identifiers in the abstract RDF model or not? Some people seem to think that what was meant to be a syntactic property of blank nodes is actually a semantic one. Etc.

One comment, there is also another scope being discussed on the RDF WG, 
essentially local to a "named graph".

answers in line from here:

> ===== BEGIN BLANK NODE QUESTIONNAIRE ===
> 
> Section 1: Syntactic Scope of Blank Nodes
> -----------------------------------------
> 
> When merging the two below RDF graphs G1 and G2 into a single graph, what would be your /preferred/ result, regardless what the current RDF spec sais (all examples given in Turtle syntax)?

it's close to trig ":G1", but that's an aside :)

> G1: {
>     ex:FoafNamedPerson rdfs:subClassOf _:x .
>     _:x rdf:type owl:Restriction .
>     _:x owl:onProperty foaf:name .
>     _:x owl:allValuesFrom xsd:string .
> }
> 
> G2: {
>     :alice foaf:knows _:x .
>     _:x rdf:type foaf:Person .
>     _:x foaf:name "Bob" .
> }
> 
> [ ] Pro graph-local blank nodes: 
>     The blank node identifiers should be /renamed/.
>     That is, after merging there will be /two/ distinct blank 
>     node identifiers, one (e.g. "_:x1") substituting the "_:x" coming 
>     from G1, and another one ("e.g. "_:x2") substituting the "_:x" 
>     from G2.

x - the above should happen

> [ ] Pro global blank nodes: 
>     The blank node identifiers should /not/ be renamed.
>     That is, the result graph will still contain
>     one single blank node identifier named "_:x".

if both were called G1 then then this option

preference: blank node identifiers are locally scoped at "named graph" 
level, where a name is known, else as per how things are now when no 
"name" is known for the graph.

Best,

Nathan

Received on Friday, 4 March 2011 16:32:52 UTC