RE: [xdi] Minutes: XDI TC Telecon Thursday 1-2:30PM PT 2010-10-14

Hi folks,

 

We're on release here sched for tomorrow morning and our 
team is down two people so everyone has to pick up slack..means I can't take the 
time for the call tonight. I'll read minutes tomorrow and send out comments 
tomorrow or over weekend.

 

Bill

From:  
[mailto:] On Behalf Of Drummond 
Reed
Sent: Thursday, October 21, 2010 1:56 PM
To: Giovanni 
Bartolomeo
Cc: OASIS - XDI TC
Subject: Re: [xdi] Minutes: 
XDI TC Telecon Thursday 1-2:30PM PT 2010-10-14

On Wed, Oct 20, 2010 at 7:27 AM, Giovanni Bartolomeo 
<> 
wrote:

Hello,

as 
  from my AP of last week, here is a message about some concerns I have 
  regarding the document

http://www.oasis-open.org/committees/download.php/39382/new-view-of-xdi-graph-model-v1.pdf

Please 
  have a focus in the main picture labelled "New View". Let's assume there is a 
  node which is the "root (context)", the next node is a "subcontext" and the 
  addressing mechanism suggests to use the statement $/=drummond/$ to address 
  drummond's context (where =drummond is a label associated to the ARC 
  connecting the root to the first hierarchical subcontext, i.e. 
  =drummond)

But issues arise when we need to describe a conventional 
  triple, =drummond/+friend/=markus. In this case, we have three distinct 
  labels, i.e. segments =drummond, +friend, =markus. Should all these be 
  associated to arcs? If this is the case, then, is it correct to say that all 
  labels are "predicates"? Actually they may represent predicates but also 
  subjects or objects.

So, could you please clarify precisely how 
  subject-predicate-object of a classical triple are represented in this new 
  view of the graph?

In thinking about this, I realized that you and Andy 
  discussed many years ago a similar idea in the ATI model (I was not member at 
  that time but I've seen some presentations on the docbox), and also that I was 
  inspired by this discussion in an earlier contribution

http://www.oasis-open.org/committees/download.php/30802/xdi-rdf-global-graph-proposal-v1.pdf

where 
  labels are all associated to arcs - not to nodes. Except for the out-of-date 
  semantics of some operators (which refers to an older version of the XDI RDF 
  model documents), maybe is it a similar view that you are 
proposing?

Giovanni, I can't tell from your document (I haven't had enough time to 
read it), but yes, the fundamental idea of applying XDI to the RDF graph model 
was that all the XRIs describe arcs between nodes, and that the XRI (and also 
the URI when you add the URI base under the XRI 3.0 rules) of any RDF subject 
node was composed of any XRI path that led to that node.

So, in the graph 
below, the self-referential arc $ identifies the root node. From an RDF 
standpoint, this means that if this node is the subject of an RDF statement, the 
URI is http://xri.net/$. That is the global 
self-context, so it's use with any other global address is optional.

From the perspective of that 
root node, =a is a predicate to a subcontext. So that yields the XDI statement 
$/=a/$ and the RDF statement <http://xri.net/$> <http://xri.net/=a> <http://xri.net/$>.

Now that node =a is 
identified, it can be referenced in an RDF standpoint as a subject node. The XRI 
of that subject node is the address from the root -- =a -- and the URI is http://xri.net/=a .

So now we can say the 
following XDI statements: =a/+friend/=b, =a/+e/"foo", and =a/+c/(+d). In RDF 
this would be:

<http://xri.net/=a> 
<http://xri.net/+friend> <http://xri.net/=b>
<http://xri.net/=a> <http://xri.net/+e> <"foo">
<http://xri.net/=a> <http://xri.net/+c> <http://xri.net/(+d)>

Hope this helps, I 
look forward to discussing on today's call.

=Drummond