RE: Comments on the new RDF Test Cases draft

>  >[on the text:]
>>
>>A parser is considered to pass the test if it produces a graph
>>isomorphic with the graph described by the N-triples output document.
>>
>>>  > This is wrong, according to the standard definition of graph
>>isomorphism
>>>  > (care when using words without accurate definitions...!).
>>
>>An RDF graph is a labelled digraph with (some) blank nodes - that is,
>>N-Triples "labels" on blank nodes are only artifacts of a graph
>>serialisation
>>mechanism. I'd always used terms like "isomorphism" to refer to the
>>appropriate equivalence relationship for the class of mathematical objects
>>I'm talking about at the time.
>
>I personally wouldn't have a problem with the original wording, but now it's
>been mentioned I suppose it has to be tied down. It's not entirely
>straightforward - ok, the labels we see might just be artifacts, but does a
>bNode carry more meaning than that suggested by the topology?

No.

>  My guess would
>be that it does (even if this will be the same for all bNodes), and so
>should be considered labelled.

Please don't consider bnodes to be labelled. We went through hell 
getting it clear that they are NOT labelled.

If you want to be achingly precise, *define* isomorphism for RDF 
graphs as follows: there is a bijection (1:1 mapping) from the bnodes 
of graph A to those of graph B which maps B to A. Then proceed as 
before. To keep rogue mathematicians at bay, say explicitly that 
isomorphism of RDF graphs is not graph isomorphism.

>
>  Would the following replacement
>>text suffice?
>>
>>A parser is considered to pass the test if it produces an RDF graph
>>(that is, a partially-labelled labelled digraph) isomorphic with the RDF
>>graph described by the N-triples output document.
>
>I'd be tempted to drop 'partially labelled'.

I tend to agree on that.

Pat

Received on Monday, 13 May 2002 11:46:43 UTC