Print

Print


From: Saašha Metsärantala
> Sent: Monday, February 07, 2011 3:18 PM
> To: [log in to unmask]
> Subject: Re: [DATETIME] Revised spec
> > 325
> We could explore the possibilities to use QName 

I think an approach based on qnames would be far more complex than the approach I have suggested, so let me describe my proposal in more detail.

The problem is expressing named events and period, which I'll refer to as "temporals" or "temporal names".

 Such as:

U.S. Revolutionary War
World War 1
Jurassic period

One could then express for example the interval "from the Revolutionary War to World War 1" combining the two expressions using interval syntax.

I propose to represent a temporal with a URI (enclosed in quotes).  

For the three examples above there do exist authoritative URIs:
http://id.loc.gov/authorities/sh85140139:   U.S. Revolutionary War
http://id.loc.gov/authorities/sh85148236:   World War 1
http://id.loc.gov/authorities/sh98002342:   Jurassic period

So the interval "from the Revolutionary War to World War 1" would be:

"http://id.loc.gov/authorities/sh85140139"/"http://id.loc.gov/authorities/sh85148236" 
 
I am not suggesting that the URIs necessarily need to identify authority records, much less temporal authority record, and they certainly don't necessarily need to use LC vocabularies. What would be required is that any vocabulary term to be used as a temporal name has to have a URI.

I suggest also that the spec strongly recommend a linked data approach, i.e. that the URI when followed leads to useful information allowing the receiver of the data to determine what the event/period is. It could lead to an RDF document, as in the following, for the temporal "16th century":

http://www.loc.gov/standards/mads/rdf/examples/sh2002012470.xml 


Admittedly, the complexity in this approach is, how do these temporals get created. Many are at LC (or will be created) and other similar vocabularies will be created elsewhere.  Otherwise, I think this is a straightforward approach, and it obviates the need to deal with QNames. 

(That is, if I understand the QName suggestion correctly: there would be a URI to identify a vocabularly, and the terms in the vocabulary would be qualified by that URI.  But I think assigning URIs directly to the term  - which could be as simple as concatenating the URI with the term - is a simpler approach.)

Please comment.

--Ray