Minutes: XDI TC Telecon Thursday 1-2PM PT 2010-02-04

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

Title: Federal Identity, Credentialing, and Access Management TFPAP, Version 1.0.1, September 4, 2009

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

Date:  Thursday, 04 February 2010 USA
Time:  1:00PM - 2:00PM Pacific Time (21:00-22:00 UTC)

ATTENDING

 

Giovanni Bartolomeo

Markus Sabadello

Drummond Reed


1) XDI RDF METAGRAPH MODEL & ALTERNATE METAMODELS

On last week’s telecon, Giovanni said that he would like to propose an alternative XDI metamodel. While Drummond supports the research into this, he has concerns about the TC pursuing more than one metamodel for the XDI 1.0 specification suite, given that with the current metamodel we are very close to finally producing the first formal XDI 1.0 specifications, and that several real world implementations badly need these specifications formalized.

 

The discussion began with Giovanni asking if the goal was to publish the 1.0 specifications while semantic issues still remained. Drummond clarified that he was talking about at least pushing forward to the Working Draft level of specification, because the TC has taken so long (6 years) just to reach that point.

 

There was a short discussion of the timing to move to Working Draft 01. Drummond said that if he is going to be the editor, he can’t do it until after the RSA Security conference the first week of March. This would give us about a month to close issues.

 

Drummond clarified that he certainly does not want to build any logical inconsistencies into XDI 1.0. Giovanni explained that his motivation for proposing a different metamodel is to define formal semantics for the XDI metamodel that help make sure logical inconsistencies are avoided. The proposal he is developing are based on the same requirements as the current metagraph model but makes some different assumptions, thus resulting in different semantics.

 

Giovanni suggests that he would like to author a paper about the XDI metamodel in order to: a) raise the visibility of the work, and b) receive the feedback of the scientific and academic community on the metamodel and formal semantics.

 

Markus observed that he has been working with implementations of the current metagraph model for over a year now and not experienced any problems, so he wondered how much value that proof of semantic correctness will add. Drummond said he pictured Markus’ question as asking about two levels: a syntactic level (“graph correctness”) and a semantic level (“logical correctness”). Consistency/correctness at the syntactic/graph level is necessary but not sufficient for consistency/correctness at the semantic/logic level.

 

However Drummond said he believes that the metagraph model will prove to be logically consistent and correct at both levels. So he is eager to continue to explore any issues of semantic correctness.

 

He cited the next topic as an example, so the focus moved to that.

 

2) $HAS SEMANTICS CONTINUED

 

Drummond said that part of the power of the metagraph model is that when used very precisely, it suggests novel solutions to challenges we have faced at modeling in XDI certain common constructs in natural language. He sent an example to the list just before the call:

 

     http://lists.oasis-open.org/archives/xdi/201002/msg00003.html

 

He reviewed the message with Giovanni and Markus to go over how a precise application of the metagraph predicate $, which identifies the current context, can be used to express both possession and plurality of XDI RDF subjects. Due to the coincidence that the $ global context symbol was chosen to represent this resource, what is truly stunning is how much the resulting XDI expression ends out reading like the same natural language expressions in English.

 

            POSSESSIVE

            =bill/$has/$                   <==>   (=bill/$)                        <==>   =bill$

            =bill$/$has/+car            <==>   (=bill$/+car)                 <==>   =bill$+car

 

            PLURAL

            +car/$has/$                  <==>   (+car/$)                        <==>   +car$

 

            POSSESSIVE + PLURAL

            =bill$/$has/+car$          <==>   (=bill$/+car$)               <==>   =bill$+car$

 

Both Markus and Giovanni laughed as hard as Drummond did when they first saw this result.

 

Markus then asked how one would render an English expression like “the car’s windshield”. Drummond pointed out that there is a significant difference between “the car’s windshield” and “car’s windshield”. The latter identifies a specific class of windshield. The former identifies one instance of that class by adding the definitive article “the”.

 

Giovanni said he has had the same issue of identifying an instance of an object that is not a person or an organization.

 

Drummond said this same issue had come up in his consideration of the English language expressions above. Ironically, the solution appears to once again to involve the key XDI concept of context. The definitive article “the” expresses that the subject being referenced is known in the current context, or more precisely, “is uniquely distinguished in the current context”. In that case, since $ represents the current context in the current XDI RDF metagraph model, and since all XDI subjects within that context are uniquely XRI-addressable, this means the XDI equivalent of the English phrase “the car” would be:

 

            $+car

 

The literal translation would be “the XDI RDF subject +car in the current context”. Note that this identifies a different XDI RDF graph than +car by itself, because +car by itself identifies the generic class of “car”, and not any specific car.

 

$+car is definitive since $ represents the current context, of which there can be only one. This means the way to express an indefinite article such as “a car” would be to make it explicit that the context being referenced could be any context. That can be done with a cross-reference, which is always relative to the XRI that it follows. So, the XDI equivalent of the English phrase “a car” would be:

 

            $(+car)

  

 

3) NEXT CALL

 

The next call is next week at the regular time. Drummond noted that due to work commitments leading up to RSA Conference the first week of March, he may have to miss several calls after next week’s call.