Minutes: XDI TC Telecon Thursday 1-2PM PT 2010-09-16

[Drummond Reed <drummond.reed@xdi.org>]

Following are the minutes of the unofficial telecon of the XDI TC at:

Date:  Thursday, 16 September 2010 USA
Time:  1:00PM - 2:00PM Pacific Time (21:00-22:00 UTC)

ATTENDING:

Giovanni Bartolomeo
Markus Sabadello
Joseph Boyle
Kaliya Hamlin 
Drummond Reed 
Joe Johnston 

GUESTS
Mike Schwartz


NOTE: THE IDEARPAD LINK FOR TODAY IS:

      http://xdi.idearpad.org/8

Please try to preface each of your comments with your name so the transcription into the minutes is easier.
(Note: the link to last week is http://xdi.idearpad.org/7 - password was "turtle". Future Idearpads will not need passwords.)


1) REVIEW OF "NEW VIEW OF THE XDI GRAPH MODEL" DOCUMENT

Drummond posted a document summarizing many of the key insights coming out of the XDI Retreat in mid-August:

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

Drummond reviewed the main sections and diagrams of this document that showed the "old view" and the "new view" and went through the implications of the new view. One of the key implications was that XDI subjects that represent new XDI contexts are functionally equivalent to RDF blank nodes becuase they are both the object of one RDF statement and the subject of another. The difference is that they have at least one unique XRI, so they can also serve as standard RDF subjects and objects.

Giovannia asked that if you have for example this statement:

=drummond/+friend/=markus

could the subject (or the object) of this statement be considered a blank node?

Drummond explained that node identified by the XRI =drummond is the object of an arc (RDF predicate) from an abstract global context root (the root of all XRIs). In that context, =drummond is an XDI object. However, as soon as =drummond is the root of XDI statements in its own context, such as =drummond/+friend/=markus, then it is an XDI subject.

At that point, the node in the graph identified by the XRI =drummond satifies the definition of an RDF blank node, i.e., it is the object of one RDF statement and at the same time the subject of another.

Another way of looking at it is that the XDI statement

    =drummond/+friend/=markus

implies the following XDI metagraph statements:

    $/=drummond/$
    =drummond/$/(*/+friend/=markus)

The first metagraph statement says that the context root $ has a subcontext identified by the predicate =drummond (which means the address of this subcontext is =drummond). The second metagraph statement says that the context =drummond is the subject of a statement in which it has a relational predicate +friend whose object is =markus.

Giovanni pointed out that in fact the XRI =drummond can be an XDI subject, XDI predicate, or XDI object. From that perspective, it is specifically when =drummond is used as a predicate that it is defining the relationship between blank nodes.

So our conclusion was that the question of whether an XDI subject node or object node is an RDF blank node depends on how you adddress it in the graph.

Drummond pointed out that this insight may provide the "missing link" to how to fully describe an XDI graph in conventional RDF triple statements. However further exploration will be needed to prove out whether this is true.

Drummond said he hopes Bill is able to read  and comment on these minutes since he is eager for Bill's feedback about  this insight.

Drummond went on to explain that another of the implications of this new view is that are three implicit "supertypes" of XDI statements: those whose objects are literals, those whose objects are references (relations), and those whose objects are subcontexts. Since $ was already being used to identify contexts, it followed that two other XRI delimters were a natural fit as identifiers for the other two supertypes: * for relations, and ! for literals.

Giovanni asked the question: what does 

=drummond/+friend/$

identify?

He said that since =drummond/+friend can have any number of objects (e.g., =markus, =bill, etc.), what specifically it means when you use the object identifier $, i.e., when you make the statement =drummond/+friend/$.

Drummond answered that the very precise semantics now proposed for $ (along with * and ! when used alone as XDI subjects or objects) means that they represent datatypes, specifically datatypes of XDI statements, which Drummond calls XDI supertypes.

 * $ represents the root of statements in any context that identify subcontexts.
 * * represents the root of statements in any context that identify relations between that context as an XDI subject and other contexts as XDI objects.
 * ! represents the root of statements in any context that identify properties of that context whose object is an XDI literal.

So Drummond's answer to Giovanni's question is that =drummond/+friend/$ is a precise metagraph statement that identifies the node in the metagraph that is the root of all XDI statements that identify the members of the set =drummond/+friend. In other words, it identifies the set of all of of Drummond's friends.

Giovanni: I cannot see this for the moment at least. To me there are only arcs labelled "+friend" which point to different nodes (=markus, =john, etc.)

Drummond said it would help to the graph using our proposed new notation to illustrate his answer. 

Note: after the call, Drummond did this and concluded that =drummond/+friend/$ is equivalent to =drummond+friend, i.e., it is an arc labelled +friend from the subject =drummond to a subcontext. The identifier of the subcontext is $, or "self". This is the XDI equivalent of an RDF blank node, since it can now be the subject of other XDI statements, however this node is in fact identified by the combination the subject (=drummond) and the predicate (+friend), so the address of the node is =drummond+friend.

This means that semantically =drummond/+friend/$ is semantically equivalent to =drummond/+friend/=drummond+friend. Another way to put this is that is it a true statement to say:

    =drummond+friend/$is/(=drummond/+friend/$)


2) POTENTIAL XDI PAPER

Giovanni mentioned a paper that he is proposing to submit.He plans to circulate a version of the paper on the list so TC members can comment.


3)  XDI JSON SCHEMA

We took another look at the current stable proposal on the JSON Format page at:

   http://wiki.oasis-open.org/xdi/JsonFormat

We still need to get Bill's feedback on this serialization proposal.


4) LINK CONTRACTS

Drummond said that an example of the core link contract pattern appears in the New View document (see link in agenda item #1 above), however we ran out of time to review it.


5) NEXT CALL

Drummond said that he must travel again for the next two weeks so is not sure he will be able to attend those calls, however hopefully Bill can chair them. Plan on another call next week.