xdi message
[Date Prev]
| [Thread Prev]
| [Thread Next]
| [Date Next]
--
[Date Index]
| [Thread Index]
| [List Home]
Subject: Minutes: XDI TC Telecon Thursday 1:00-2:00PM PT 2011-06-16
- From: Drummond Reed <drummond.reed@xdi.org>
- To: OASIS - XDI TC <xdi@lists.oasis-open.org>
- Date: Thu, 16 Jun 2011 16:35:07 -0700
Following are the minutes of the unofficial telecon of the XDI TC at:
Date: Thursday, 16 June 2011 USA
Time: 1:00PM - 2:00PM Pacific Time (20:00-21:30 UTC)
ATTENDANCE
Mike Schwartz
Bill Barnhill
Giovanni Bartolomeo
Markus Sabadello
Drummond Reed
Joseph Boyle
THE IDEARPAD LINK FOR TODAY IS:
1) GIOVANNI UPDATE
Giovanni reported on the presentation of his paper at the European Semantic
Web conference. He also had a poster about a way of integrating XDI within
the Linked Data project.
The
first issue was explaining that XRIs as abstract identifiers can be
bound to concrete URIs, so they can be compatible. We can also
accommodate URIs as cross-references. So there is no incompatability.
Giovanni
talked to Prof. Chris Bizer, Free University of Berlin, inventor of
DBPedia, who started the Linked Data project in 2005. The Linked Data
project now has over 18 billion triples (!).
Giovanni
talked to him both about XDI graph addressability and also the
owl:sameAs problem. In particular, XDI can help address the issue of
using the same URI in different contexts (i.e., different RDF graphs).
He
also had discussions about whether RDF reification can be used to do
the same thing as XDI graph addressability. Giovanni pointed out that
while in theory it is possible, it requires 5X the number of URIs, and
is computationally much more expensive.
Giovanni
also showed the OxGraph demo. The perception of the audience was that
the use of symbols was hard to understand vs. the mnemonic terms used by
RDF/OWL.
In
conclusion, we now have a way to translate XRIs and XDI documents into
URIs and RDF documents that is compatible with Linked Data. The
challenge is to communicate the advantages clearly. Giovanni believes
the key advantage is provenance, i.e., being able to express extend the
XDI concept of context to a URI.
Bill
summarized the problem this way: in RDF, URI namespace are independent
of RDF graphs, while in XDI, the namespace is directly related to each
XDI graph. Currently the solution for adding this type of context in RDF
is Named Graphs, however this requires going to a quad instead of a
triple, and the actual triple is still not itself contextualized. So
Giovanni's suggestion is to use multi subsegmented identifiers such as XRIs to merge the URI name of the graph with the URI of the subject (or also predicate) of each triple so that contextualization becomes built into the RDF graph itself, as it happens in XDI documents.
Bill: A page that may provide some extra information related to this is...
2) MOVING BACK TO SEMANTIC $WORDS INSTEAD OF SYMBOLS
As
the feedback Giovanni received shows, there is a growing sense that the
value of using pure symbols for the XDI metagraph relationships is
offset by the readability of semantic $words. Even Drummond, who led the
charge towards adopting symbols after the suggestion from Eve Maler, is
leaning back in the other direction.
# DRUMMOND to investigate moving back to $words instead symbols and make a proposal.
3) OPENXDI STATUS REPORT
Mike
reported that messaging is done and basic link contracts is done.
Authentication is the next step. The project is leaning toward's Bill's
suggestion of implementing PKI first. The main reason is that it is the
most granular. Initially the keys will be on the server, but stronger
key management can be added at any time. SAML and OAuth can be added
afterwards.
He
said that 6/30 will be the first official 0.0 build. Mike then went
over the upcoming development schedule, including adding Messaging to
OxGraph, and completing link contracts.
4) UPDATED LINK CONTRACT PATTERN
Mike
asked for some additional features to be demonstrated in the link
contract pattern. Drummond included them in the latest version of the
XDI Graph Patterns doc:
Mike
would also like examples of $not$xxx permissions, i.e., deny instead
of allow. Drummond has not had time to illustrate this yet.
5) XDI 1.0 DICTIONARY
The following wiki page was been refactored per our discussion a few weeks ago:
Of
these specs, Mike would like to make progress fastest on the
Dictionary spec. The proposal is to start a wiki page,
XdiOneDictionary, which would use either XDI statement format or JSON
format to express the initial XDI dictionary.
# DRUMMOND to create this page.
Bill added the following index of what he's creating from DITA and plans to post by this weekend.
Concept: An XDI Message definition
Concept: Definition of a trustable XDI message
Concept: Characteristics of a trustable XDI message
Concept: Parts of an XDI Message
Concept: XDI Message exchange patterns
Concept: Definition of data interchange
Task: Creating an XDI Message
Task: Creating a Trustable XDI Message
Concept: Definition of a Link Contract
Concept: Link Contract example
Concept:: Connection-less XDI Messaging
Concept: Connection-oriented XDI Messaging
Task: Establishing an untrusted XDI Connection
Task: Upgrading an untrusted XDI Connection to be Trusted
Task: Pt. 1 of upgrading to trusted: authentication (PKI first, OAUTH to come later)
Task: Closing an XDI Connection
Task: Negotiating a new Link Contract
Task: Verifying an XDI Connection upgrade request
Concept: The XDI core operations
Concept: XDI Get Operation
Concept: XDI Set Operation
Concept: XDI Modify operation
Concept: XDI Delete operation
Concept: XDI Subscribe operation, for pub-sub
Concept: XDI Take operation, for knowledge spaces (L-Spaces)
Concept: XDI Put operation, for knowledge spaces (L-spaces)
Concept: Definition of an XDI Dictionary
Concept: The prime XDI Dictionary
Concept: Querying the prime XDI Dictionary
Concept: Creating your own dictionary
Task: Create your own dictionary
Concept: The locations of dictionaries applicable to an XRI
Task: Constructing the addresses of possible dictionaries
Concept: Basic XDI types in the prime dictionary
Concept: The core XDI relation types
Concept: XDI relation type hierarchy
Concept: XDI equivalence
Concept: XDI composition
Concept: XDI qualities
Concept: The XDI Constructors for the core relation types
Concept: XDI Constructor example
Concept: $msg
Concept: $json
Concept: $xml
Concept: $pki
# BILL to post his dictionary proposal document.
6) BILL'S QUESTIONS
We
agreed that the list of questions Bill sent to the list should be
revisited after we decide about the metagraph symbols issue. For
reference, Bill's message is at:
For convenience, and notetaking, these questions are:
Q1:At
different times we've used a different dollar word to indicate that
something is an instance of something else, an 'is-a' relationship. I
believe $ at the end of the subject with a * predicate is currently used
to represent the inverse of is-a, is that correct? By that I mean that
the sentence "Bob's ball 1 is an instance of an Example org ball" is
stated by @example+ball$/*/=bob+ball$1. Does this match everyone's view
of consensus? If that's not currently accepted way then what is would be
the accepted XDI statement?
Q2:
If I'm Bob let's say I'm the data authority for Bob's ball 1, but I may
or may not be the authority for @example+ball. Let's assume I'm not.
Then what is the currently accepted way to state
@example+ball$/*/=bob+ball$1 using an inverse so the XDI statement is
under =bob? Is it =bob+ball$1/$!/@example+ball ?
The individual patterns are listed below:
.. The predicate * means that the object is a part of the whole that is the subject (see below for some questions on use of *)
.. An predicate of $! means the subject is an instance of the object: {s}/$!/{o}
.. An ending $ in a subject means the subject is a type. {s}$ <==> {s}/$!/$
..
There is no specific predicate that is the inverse of $!, but the
inverse can be specified through the rule {s}/$!/{o} <==>
{o}$/*/{s}
.. Curly braces are usable within an XDI XRI and represent a variable
Is there anyone who has a different view?
Also,
as I understand it in XRI the * character represents re-assignable
delegation. Can we use * as a segment by itself like this without
causing conformant XRI parsers problems? Also is the predicate * above
actually * or **?
Finally,
what are the thoughts on the following new dollar words to address
relationships where the subject is in a specific type of context of the
object:
{s}/$what/{o} - A more readable single predicate form of {s}/$!/{o}, puts subject in type context of object
{s}/$when/{o}
- Puts subject within the temporal context of object. Object would most
likely be a time, time range, etc. but could also be something other
than a time or time range which would mean the subject had the same
temporal context (i.e. lifetime) of the object.
{s}/$where/{o} - Same as $when, except for geographic location rather than temporal
{s}/$who/{o}
- Puts subject within context of an agent. This could mean ownership,
the actor if subject is an event, or just that subject is associated
with the agent in some way
{s}/$why/{o}
- Puts the subject, which must be an XDI statement or graph, in the
context of an object that is a justification for the subject (not
necessarily the only justification).
The $why one needs an example and I am working on it. Will post to wiki.
6) NEXT CALL
The next call is next week at the regular time. Agreed subjects:
- Review of Core 1.0 specs and path forward with each.
------------
ONGOING ISSUES LIST
Each of these is a candidate for the agenda for future calls.
* TRANSACTIONAL INTEGRITY FOR XDI (added 2011-03-24)
Since
versioning, as one example, involves multiple transactions that must
be commited as a group, we will need to address transactional
integrity. Specifically, we need to define how this will be handled at
the protocol level, vs. the implementation level.
* PROPOSED CONSTRUCTS/OPERATORS FOR XDI
Discuss the following wiki page originally posted by Giovanni:
* DICTIONARY STRUCTURE
Mike would like an example of the PDX dictionary as soon as we can do it.
* EQUIVALENCE SEMANTICS
Close on whether we need an additional $ word that is the equivalent
of Higgins Personal Data Model (PDM) semantics of h:correlation,
which is not as strong as $is.
* COOL URIS
Continue previous discussion about the use of standard RDF URIs in XDI:
[Date Prev]
| [Thread Prev]
| [Thread Next]
| [Date Next]
--
[Date Index]
| [Thread Index]
| [List Home]