Re: Making progress on graphs

On 05/13/2012 04:25 PM, Sandro Hawke wrote:
> On Sun, 2012-05-13 at 15:20 -0400, Peter F. Patel-Schneider wrote:
>> On 05/13/2012 02:15 PM, Sandro Hawke wrote:
>>> On Sun, 2012-05-13 at 13:36 -0400, Peter F. Patel-Schneider wrote:
>>>> The syntax part of Sandro's document appears mostly fine to me.
>>> Great.
>>>
>>>> However, there is quite a bit more to Sandro's document than there is to
>>>> Richard's proposal.  I'm not in favour of the semantics and
>>> As long as an RDF graph has truth conditions, I think a Dataset also has
>>> to have truth conditions.   Do you agree with that?
>> No, and I have so stated in the past.
> I suppose you have, yes.   Can you explain why RDF should have formal
> semantics?   (I think this is the same as saying, "... should have truth
> conditions"; if I'm wrong about that, please correct me.)
>
>     -- Sandro
>
>

Semantics is not the same as truth conditions, witness databases.

I think that RDF needs to have a formal semantics, because there is supposed 
to be consensus on what the constructs in RDF mean, and thus what follows from 
particular RDF (mostly S) graphs.

My understanding is that the semantics for RDF came about because although 
there was consensus on the general meaning of RDF pre-2004 but there were 
misunderstandings about the details, leading to differing implementations.  
There were also some bits of RDFS where the informal meaning was not tenable, 
and so ended up being changed.  This lead to the further understanding that 
there were parts of RDF(S) where there was no consensus concerning the meaning 
of the constructs, and these constructs have ended up being rather vestigial.

peter

Received on Sunday, 13 May 2012 22:18:35 UTC