On 11/4/14 11:20 AM, Simon Spero wrote:
> Some comments-
> Technical points:
>  1. Anyone can create a subclass. A subclass does not have to be in
>     the same namespace as the original class.
>  2. If a system encounters an instance of a class that it has no
>     specific knowledge of,  it can handle it  as an instance of one of
>     the class's  superclasses.
>  3. If a system encounters a string that it has no specific knowledge
>     of, it has a string.
>  4. If several subclasses are defined in different places that have
>     the same meaning, they can be declared to be equivalent classes.
>  5. Metadata can be attached to classes;  this can include labels,
>     identifiers, and display hints.
>  6. It is possible to define classes as being things that have a
>     specific string as the value of a property. This can be used to
>     infer an instance's class given the property value, or the
>     property value given the class.
>  7. It is possible to specify Key properties for a class, so that any
>     two URIs which have the same values for all of those properties
>     can be inferred to be two different names for the same thing.
>  8. rdf:value has a range of rdfs:Resource.  It is thus a source of
>     URI/String punning.

Simon, sorry but this is pretty much irrelevant to what I was saying, 
and definitely contains some explaining that I consider to be 
unnecessary in this context. I am not questioning the flexibility of RDF 
as a technology, and don't think that can be construed from my 
statements. I *am* questioning whether the library standards and 
technology environment will support these capabilities. If they don't, 
then it doesn't matter what the possibilities of RDF are. How long will 
we have to wait to have a vendor-supplied system that that does any of 
your 8 points, above? You state them as technical givens, but at the 
moment there is, AFAIK, exactly zero library systems that function in 
this way (and I'm including systems in RDF like Europeana and DPLA). I 
may be extra cynical (not to mention a bit old), but I'd be surprised to 
see these included in library systems in my lifetime.


> Policy points.
>  1. If only there were some sort of precedent for setting up some sort
>     of  program for cooperative cataloging that could let
>     appropriately trained library staff create values that could
>     somehow be turned into linked data objects... some sort of type
>     authority cooperative program (TACO).

Karen Coyle
[log in to unmask]
m: +1-510-435-8234
skype: kcoylenet/+1-510-984-3600