Re: [Handle-globus] The LDAP URL Format

[Frank Siebenlist <franks@mcs.anl.gov>]

Sam X. Sun wrote:

> ....
>
>>
>>> I think the mapping should work for our purpose, where AA will know
>>> which handle prefix to use. On the other hand, Itmight help to point
>>> out that this is not a general mapping between the URI/DN and handle
>>> as I originally thought. Instead, this is a mapping from URI/DN to
>>> handles under a specific handle prefix.
>>
>>
>>
>> Initially, I was looking for a single way to map those unique
>> identifiers, like DNs, into a single way into the handle name space. You
>> could maybe/probably find a scheme to do that by dedicating a single NA
>> to map certain URI schemas in.
>>
>> However, we're not issuing the URIs as they already exist, but merely
>> binding attributes to those names.
>> Now, theoretically anyone in the world could potentially bind any
>> attribute to any unique name, and it becomes an issue that you only will
>> be interested in certain authorities to do that. It this case, a handle
>> system Naming Authority is not so much a keeper of the name, but a
>> keeper of certain name bindings that have a trust and application domain
>> as indicated by that NA.
>>
>> I'm still struggling to find the right mental model here...
>>
>
> I agree with your points here... likewise, still need some thinking on
> the overall picture


I just reread the RDF-primer.
http://www.w3.org/TR/rdf-primer/

I do not want to suggest that we should use the handle system to build a
complete RDF-database, but the general concept of having URI as
identifier for things/subjects, with again URIs to identify
predicates/properties of those things, where the property-values/objects
are either again URIs or (typed-)literals, give you a nice general
framework to work in. (To be more precize, RDF uses URIrefs, but that's
a minor thing.)

Now every URI could be mapped to or could be itself a handle, where the
handle-value-type would be another URI specifying the predicate/property
(which could again be associated with a handle), and the handle-value
(object value) itself would either be a literal or another URI (with a
possible handle associated with it).

Still a little unclear how to add the xml-schema datatype to the
handle-value... maybe it should be a property of the property, i.e.
maybe we should have a handle-value for the property-URI-handle that
specifies the datatype of the object... there are also example that keep
the type-identifier with the object.

Type-identifiers are again URIs refering to type specifications.

All those URI get very long and the use of namespaces identifiers could
be benficial... maybe we could use a convention where a handle-value of
the "12.34/uri:" handle would hold the abbreviations for the full
namespace identifier that can be used for all the URIs that are mapped
underneath (?).

One of the things lacking in RDF is the concept of the issuer, i.e. the
sayer or authority, of the statements... I have to browse some more
because I vaguely remember there were some docs that tried to model that
concept into the RDF-framework.

Keep dreaming...
-Frank.


-- 
Frank Siebenlist               franks@mcs.anl.gov
The Globus Alliance - Argonne National Laboratory


_______________________________________________
Handle-globus mailing list
Handle-globus@cnri.reston.va.us
http://www.cnri.reston.va.us/mailman/listinfo/handle-globus