Re: Graph terminology in spec text.

On 21 July 2011 09:27, Andy Seaborne <andy.seaborne@epimorphics.com> wrote:
> On 21/07/11 01:08, Ian Davis wrote:
>>
>> We recognise that "graph" has subtly different semantics
>> between sparql and rdf concepts so let's avoid that term.
>
> It would be helpful to if you could point to text in the SPARQL specs that
> leads to that conclusion.  It should be changed because a graph is value and
> used as such in SPARQL, both in query and update.  As such , the changes are
> editorial and not a cause for a SPARQL second last call.

If the terminology can be unified and 'graph' kept in our technical
vocabulary, there are significant benefits for RDF.

The word 'triple' seems to alienate many people, whereas the wider
tech scene is lately abuzz with all kinds of talk of 'graphs':

 - the social graph
 - interest graphs
 - graph databases

I appreciate the importance of being technically unambiguous, ... but
let's not talk ourselves into obscurity here! The fact that this
notion of 'graph' is being widely spread really helps our project, so
please don't abandon the word just when it is becoming more widely
appreciated.

In terms of outreach, what's that difference between 'graph' and
'triple'? Mainly that 'graph' emphasises that the first and third
pieces of each triple tend - on a good day - to draw from the same
pool of potential values, and can consequently connect up to form a
linked data structure. This is a big big part of the appeal of RDF,
with or without Linked Data's clever tying of this idea to the notion
of 'the Web graph' of linked documents. In some sense, the Semantic
Web project is about creating the graph of all data graphs; I'm
willing to forgive a little inclarity for the rhetorical benefits that
come with the word.

cheers,

Dan

Received on Thursday, 21 July 2011 08:40:39 UTC