Re: Labelled graphs

On Wed, 2012-04-25 at 14:06 +0100, Andy Seaborne wrote:

[taken out of order]

> And there are other things in the charter.

Yes, indeed.   Many of us keep saying that other things need to take
priority, but perhaps we're not saying that loudly enough, and perhaps
our actions (sending lots of email about graphs) speak even more loudly.

I would support a plan to fork graphs off onto another meeting time and
even another mailing list until there is consensus among the people
participating in that alternative forum.


> On 25/04/12 13:46, Sandro Hawke wrote:
> > On Wed, 2012-04-25 at 13:11 +0100, Andy Seaborne wrote:
> >>
> >> On 25/04/12 12:28, Ivan Herman wrote:
> >>>
> >>> On Apr 25, 2012, at 05:16 , Sandro Hawke wrote:
> >>>
> >>>> On Tue, 2012-04-24 at 16:05 +0100, Andy Seaborne wrote:
> >>>>>
> >>>>> On 24/04/12 13:04, Sandro Hawke wrote:
> >>>>>> (mostly agreement, a few details)
> >>>>>>
> >>>>>> On Tue, 2012-04-24 at 12:03 +0100, Andy Seaborne wrote:
> >>>>>>>
> >>>>>>> On 17/04/12 16:59, Guus Schreiber wrote:
> >>>>>>> ...
> >>>>>>>
> >>>>>>>> An attempt at formulating a possible conclusion/consensus from this thread:
> >>>>>>>>
> >>>>>>>> * Non-typed labels are simply associations, no special semantics
> >>>>>>
> >>>>>> There are some semantics, though: the label IRI (or blank node) denotes
> >>>>>> something (maybe call it a "labeling object"), and that something is
> >>>>>> associated with the graph.
> >>>>>
> >>>>> Given the "something" indirection, whether that counts as "semantics" or
> >>>>> not is a bit moot to me.  It's "no fixed semantics".
> >>>>
> >>>> Here's the part that's important to me:
> >>>>
> >>>>          Under OWL entailment and our dataset semantics, does
> >>>>             {<u1>   owl:sameAs<u2>   }
> >>>>             <u1>   {<a>   <b>   <c>   }
> >>>>          entail
> >>>>             <u2>   {<a>   <b>   <c>   }
> >>>>          ?
> >>>>
> >>>
> >>> FWIW, in the case of http://www.w3.org/2011/rdf-wg/wiki/Graphs_Design_6.1/Sem the answer is clearly yes.
> >>
> >> I think this is at a different level.
> >>
> >> (I think that) the most basic thing we (this WG) needs to do is define
> >> syntax so that things do not go wrong.  This does not need to fix the
> >> semantics for all time.
> >
> > To be clear, are you saying you would like the entirety of this WG's
> > output concerning "named graphs" to be a Recommendation for the grammar
> > of TriG?
> 
> No.  I'm suggesting we must at least provide that.  It would be a step 
> forward.

Agreed.

> Given the discussions, I am not convinced that fixing one semantics is 
> desirable.  I see different viewpoints, from different use cases, all of 
> which have merit.  Deciding the semantics now feels like research, there 
> not being one as input.

To me, it feels like what we did in making RDF 2004.   

(Of course, you could say that was a mistake, but it's hard to know
where we'd be otherwise.)

The problem is, I think, that the only way to find out if some design
works is to convince a lot of people to try it at the same time.   How
can you do that?  It looks to me like the way we do that in the Semantic
Web community is to make it a W3C Recommendation (or a least Candidate
Recommendation).   

I suppose another approach would be to have a back-room chat amongst the
people behind some key SemWeb tools (eg Jena).   But is that better than
the W3C process?   We could set a very high bar for CR on this, if we
want to make sure it doesn't get to Rec until "it actually works". 

> I think the best approach is to define the absolute minimum, and publish 
> the semantics as based on that.  This means if the semantic proposals 
> are found wanting (not in a technical sense, but in a utility sense c.f. 
> reification) others can also emerge.
> 
> This isn't giving up - it's realising there is a minimum and 
> possibilities on top of the core basics.
> 
>  > I note that our charter says:
>  >
>  >          Required features
>  >                * Standardize a model and semantics for multiple graphs
>  >                  and graphs stores
> 
> My conclusion currently is that there are different ways of using named 
> graphs in applications and they are all valuable.  The use cases are not 
> met by some common semantics choice directly.
> 
> Even if they can all be built on one common semantics (and this itself 
> is not obvious to me - I can see different strands still running), it 
> can still end up too complicated in practice.  So we can meet the letter 
> of the charter and still not improve the situation.

Among the people who want to work on this, I'd like to see if we have
consensus on a few simple bits, like the sameAs entailment test earlier
in this thread, or the idea that the default graph is asserted.
Maybe we can do that today, or maybe today's telecon can be spent on
other things we need to do.   

    -- Sandro

Received on Wednesday, 25 April 2012 13:48:36 UTC