XDI TC Minutes

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

Date: Friday, 6 March 2015 USA
Time: 09:00AM - 10:30AM Pacific Time (16:00-17:30 UTC)

ATTENDING

Ning Zhang
Drummond Reed
Markus Sabadello
Joseph Boyle
Phil Windley
Les Chasen
Peter Davis

PRESENTATIONS/DISCUSSIONS

Review of “The XDI Graph Model” Section of XDI Core 1.0

Drummond has been working all week on the new XDI Graph Model section of XDI Core with TC members. You can see that section in the outline:

https://wiki.oasis-open.org/xdi/CoreSpecOutline

Drummond explained his work on the XDI Graph Model section, specifically “Overview” “Context Types”, and “Context Functions”. This section explains that every context node has a “context symbol” and a “context function”. There are “primary” and “secondary” context functions. Every context node must have a primary context function (root, entity, attribute), and may have one or more secondary context functions (collection, variable, definition).

Phil suggested that the term “context function” was confusing. Markus suggested the term “context modifier”. Phil felt that the term “context role” would be most intuitive to developers. Drummond pointed out that the term “role” already had a particular meaning in this space—description logic—but that the terms used in the space were already mixed. As evidence he pointed to the following section of the Wikipedia article on description logic:

*********************************

Nomenclature

Differences from First-Order Logic

The description logic (DL) community uses different terminology than the first-order predicate logic (FOL) community for operationally-equivalent notions; some examples are given here:

FOL	DL
class	concept
property or predicate	role
object	individual

The Web Ontology Language [OWL] mostly uses FOL terminology, in spite of being an implementation of a description logic.

***********************************

However after discussing the options, the TC members unanimously agreed with Phil that the term “context role” was better than “context function”.

#CONSENSUS: We will use the term “context role”.

#ACTION ITEM: DRUMMOND to change “context function” to “context role” in the next XDI Core draft.

Joseph said he was working on a chart that shows which context symbols and context roles can be combined.

#ACTION ITEM: JOSEPH to complete his table for the next draft of XDI Core.

Star Shift, Immutability Symbol, and Xref Proposal

As Drummond and Joseph and Markus work on the ABNF for the Core spec, they are considering a new way of standardizing immutability that they want to discuss with all TC members. This proposal is documented here: https://wiki.oasis-open.org/xdi/StarShift

It consists of three parts:

The context symbols @ and * are used to identify instances (individuals in description logic) that can occur either inside or outside a collection.
Introduce the immutability symbol ! that can optionally appear after a context symbol.
Change “cross-reference” to “external reference”, which can contain a local identifier or a URI, but no longer an XDI address.

Markus’ first reaction was that this felt unfamiliar, but that he was not opposed to the proposal.

Joseph said it was a step in the right direction. Joseph also suggested that as a convention integers should always be considered mutable, and UUIDs immutable. Drummond replied that by default all XDI identifiers except classes f($ and #) are reassignable (mutable), and that therefore the ! immutability symbol is needed to indicate persistence (immutability).

Phil said he was in favor of the overall proposal, especially the simplification to six context symbols, and mentioned specifically that he had always thought of ! as meaning immutability.

Drummond pointed out another advantage of this proposal: under the current syntax, immutable XDI addresses for individuals or organizations (“cloud numbers”) always used to require two context nodes, since it was not possible to express immutability outside of collections. This was different from a mutable XDI address for a person or organization (“cloud name”), which as an entity singleton only required one context node. Under this proposal both cloud numbers and cloud names would require only one node.

By the conclusion of the discussion it was agreed that the XDI Core editors should proceed to incorporate this proposal into the next draft of XDI Core.

XDI as an “Extension” of RDF

Drummond pointed out that he has added a section to the Introduction of XDI Core called “How XDI Builds on RDF”. He proposes that this section just be a short synopsis of the paper the TC has in progress on this topic, and that this section make an informative reference to that paper. There was consensus that this was a good approach.

We further discussed the relationship between XDI and RDF. Drummond showed an academic paper about RDF blank nodes which he found as he researched the topic in preparation for the “How XDI Builds on RDF” paper. We talked about how RDF blank nodes express a certain kind of “context” similar (but not equivalent) to context in XDI.

Peter said the statements about blank nodes that Drummond has included in The XDI Graph Model section that we reviewed on the call are accurate, but that the subject is so complex (as the academic paper illustrates) that it would be best to concentrate on a clear definition of XDI context nodes and contextual statements.

Markus raised the question of how the fact that XDI “contexts” do not have a direct equivalent in RDF will be explained in the XDI Core spec. He said that in his experience, people familiar with RDF often consider the tree-like structure of XDI contextual arcs a constraint until they understand that contextual arcs exist in addition to XDI relational arcs, and that relational arcs can be used the same way in RDF. So, with the exception of locally-identified blank nodes, the XDI graph model is a superset of the RDF graph model, and XDI contexts are an additional feature rather than a constraint.

Drummond agreed and will make sure this is covered sufficiently in the section of the XDI Core spec that explains how XDI aligns with Web and RDF architecture.

Holding a Committee Specification Draft Vote on XDI Core 1.0

We have talked since January about the goal of having XDI Core advance to a Committee Draft (technically a “Committee Specification Draft” in OASIS terminology) by IIW. Rather than needing a week to complete a vote in advance of IIW, Phil suggested we schedule and hold an official XDI TC meeting at IIW and approve it there. We agreed the best time was after the demo period on Wednesday April 8.

Everyone agreed that was a good idea.

#ACTION ITEM: DRUMMOND AND MARKUS to schedule this meeting with Chet Ensign.

NEXT CALL

The next call is next week at the regular time.

xdi message