Minutes: XDI TC Telecon Thursday 1-2PM PT 2009-03-19

["Drummond Reed" <drummond.reed@cordance.net>]

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

Date:  Thursday, 19 March 2009 USA
Time:  1:00PM - 2:00PM Pacific Time (21:00-22:00 UTC)

ATTENDING
John Bradley 
Giovanni Bartolomeo
Markus Sabadello
Drummond Reed

REGRETS
Bill Barnhill
Nick Nicholas



1) CONTINUE $HAS AND +X/+Y/+X+Y DISCUSSIONS

See the minutes of the second-to-last last meeting at:

	[1] http://lists.oasis-open.org/archives/xdi/200903/msg00005.html 

Also the most recent posts at:

	[2] http://lists.oasis-open.org/archives/xdi/200903/msg00004.html
	[3] http://lists.oasis-open.org/archives/xdi/200903/msg00006.html
	[4] http://lists.oasis-open.org/archives/xdi/200903/msg00007.html
	[5] http://lists.oasis-open.org/archives/xdi/200903/msg00008.html

We continued the discussion by reviewing the new wiki page posted by
Giovanni:

	[6] http://wiki.oasis-open.org/xdi/XdiMetamodel


Giovanni started by saying that =abraham/+son/=isaac belongs to the
class/instance level. By contrast the metagraph statement =abraham/$has/+son
describes that an arc labelled +son is drawn from the node =abraham.

He explained tha tif two terms are subject and object of a statement
containing the predicate $has, then a new node is introduced in the graph,
whose identifier is the composition of the subject's identifier and the
object's identifier. In this case, the new introduced node is =abraham+son

He noted that =abraham+son is a node, whereas =abraham/+son identifies the
set of nodes which are the objects of the predicate =abraham/+son (as
described in XDI/RDF v12).

However, =abraham/$has/+son does not tell anything about the relation
between =abraham and +son. To specify that this relation is actually an
aggretion, the predicate $has$a is used: =abraham/$has$a/+son. $has$a gives
you more information than $has. It says that the outgoing arc is related to
the node in a aggregation relationship.

Giovanni proposed that one way to interpret this is that $has could
specialize into $has$a, $a, $is$a, and others. Drummond noted that the only
issue here is that the other metagraph predicates do not necessarily imply a
$has arc - $has has its own particular specialization (addressing).

Giovanni said another rule is that every XDI/RDF predicate (either in the
model or in the meta-model) defines, in the metamodel, the group of its own
objects. For example, with regard to =abraham/+son/=isaac, which infers
+son/$a/=isaac, tells that the group +son, has =isaac among its members.

Giovanni suggests that the terms "group" and "member" are more accurate in
the metalevel than "class" and "instance" or "individual" when it comes to
XDI RDF.

Giovanni it is possible to further identify which subgroup =isaac is member
of =abraham+son/$a/=isaac tells that the group =abraham+son, has =isaac
among its members (the only new thing here is the terminology).

Giovanni next said $a is a special case of $has. He concludes that having
assigned the metaword $has the semantics of describing the connection
between a node and any generic outgoing arc the node may own, a statement
like =abraham+son/$has/=isaac and an identifier like =abraham+son=isaac do
not describe anything about the relationship between =abraham+son and
=isaac, apart that they are connected through an arc.

Drummond said he wants to explore this further because of the inherent
predicate semantics of any RDF arc, i.e., do all RDF arcs represent XDI RDF
$is$a statements. This is an open issue.

Giovanni said =abraham+son=isaac describes =isaac in the context of
=abhrahm+son. This asserts that =abraham+son/$a/=isaac (of which the inverse
is =isaac/$is$a/=abraham+son). If $is$a is also a special case of $has, that
would mean =isaac/$has/=abraham+son.

Drummond said that the relationship between the two statements is that
=abraham+son/$has/=isaac infers that =abraham+son/$a/=isaac, but the inverse
relationship between the two statements is not required to be true.

Giovanni said Nick suggested $is$a could be a special case of $has like
$has$a is a special case of $has. Drummond believes it is the other way
around: $has is actually a special case of $has$a. That's why
=abraham+son/$a/=isaac and  =abraham+son/$has/=isaac are different
statements. The first expresses a subtyping relationship. The second
expresses an addressing relationship that, by our earlier discussions,
_also_ represents a subtyping relationship. So we could say that a $has
relationship infers a $a relationship, but a $a relationship does not infer
a $has relationship.

Giovanni said $has could be generic (Nick calls it "vague") and could be
used to describe the relationship between nodes linked by any other $
predicates. Drummond said $has actually creates a very specific type of
relationship between two nodes: an addressing arc, the concept of which is
not part of standard RDF. 

John said that =abraham/$has/+son creates a class (what Giovanni calles a
"group") representing an aggregation, and =abraham+son/$has$a/=isaac assigns
an instance to that aggregation. So $has is about classes and $has$a is
about instances.

Drummond said that $has has one "specific" thing about it, which is that it
create the addressing relationship between the nodes.

Giovanni said that he thought that was true, but in reading Nick's email he
realized something different.

Unfortunately at that point we ran out of time, so we will try to continue
this discussion on the list, or worst case next week's telecon (provided
Giovanni can make it).