PROPOSAL schema.openid.net for AX (and other extensions)

[Mark Wahl]

Rowan Kerr wrote:

> That's not a bad list, although there are some attributes missing
> that I would have expected to see given all the sources used to
> compile the list. Birthday? State/Province, Country?

Birthday isn't in the set.  None of the three _original_ driving
applications for X.500/LDAP schema would have needed it:

ITU: support searching and browsing of "white pages" listing of
      residential and business telecoms subscribers
ISO: support lookup of OSI presentation address of an application
      entity
IETF: support search of employees in research network and intranet
       "white pages" services

Some of the Project Liberty schemas might have included birthday,
but I haven't had a chance to look at them recently.

State or province name is there, it's called "st", and
Country name is "c".

> Not to mention I clicked through about 4 different URLs
> and still couldn't figure out what the data for a particular
> attribute was supposed to look like :) .. everything is
> a owl#Class type? (I'm pretty new to this RDF stuff so
> maybe I just haven't learned to read the docs properly).

The data in the RDF files is very limited at this time, as we
haven't yet decided on what metadata is worth incorporating,
and how best to represent it.  Yes, everything is a class.

The Java 'schemat consumer' implementation I mentioned on the
OpenID general list is an attempt to address this issue: it
implements an RDF parser and generates XHTML from it summarizing
the metadata for an attribute using normal microformat-style
HTML presentation techniques like DIV and SPAN.  I'd imagine the
Higgins project are considering something similar that presents
the data in an Eclipse UI.

> The main difference I'm seeing is that AX metadata uses
> rdf:Description while you used owl:DatatypeProperty
> and owl:Class? And the rdf:type values for AX metadata
> use the w3 XMLSchema types...

The Datatype property wouldn't be needed by OpenID.  That it is
an OWL class means it is a specific kind of rdf:Description.  My
schemat consumer implementation doesn't require that the metadata
for an attribute be an OWL class, but it helps with other
research activities which use RDF to describe metadirectory-like
transformation from one protocol/schema to another.

> If some ID attributes were added to the elements, it would be
> easier to navigate the document, as the fragment would resolve
> to something useful in a browser.
> 
> Would you consider filling in the rdfs:comment elements and
> adding openid:example elements? 

Yes, I would like to do this.

Mark Wahl
Informed Control Inc.