On 7/11/14, 8:58 AM, Ford, Kevin wrote:
[log in to unmask]" type="cite">
Important note: As easy as it is to write that individual libraries can 
and should mint their own properties when they believe they need to and 
as easy as it is to understand that idea conceptually (and I do believe 
this is the way to go), there is nevertheless a very real education 
initiative needed to diffuse this knowledge.  Education about this would 
not only help individual librarians understand /how/ custom "relators" 
can be achieved but would also be able to educate those librarians about 
how best to publish information about those custom properties in such a 
way that others can decipher the data.  Also, if this is understood by 
individual library organizations, then when those libraries interface 
with their vendors, those libraries are well-informed to inquire about 
whether the software the vendor is offering is capable of such a thing. 

Just as today's database management systems "mint" internal identifiers, with little or no prompting from their users, it would not seem to be unreasonable for those systems to mint "http" identifiers as needed for each new value in the database. I think the question comes down to whether we expect to expose those identifiers in a public way at the time of creation, but if I understand Rob's point, I don't think that's an issue. The main thing is that by requiring identifiers we can greatly simplify the data model.

As libraries export and share their data, those identifiers would become visible. Knowing the library world, we'd probably want a standard so that systems could act separately on the "ad hoc" values. Note that the "http-ness" could be a function of the export, depending on how the system stores data internally. I'm not sure that we must assume full http URIs internal to systems any more than we assume full MARC records. As long as the system can export valid RDF, how it handles its internal storage isn't relevant.

As for the "will the vendors get on board?" question, if they get on board with RDF, then this naturally follows. If they do not, then the question is moot.

That doesn't mean that we don't need to educate our community about identifiers (http or otherwise). But I expect URIs to be mostly invisible to catalogers and catalog users, in the same way that internal DBMS identifiers are today.


Karen Coyle
[log in to unmask] http://kcoyle.net
m: 1-510-435-8234
skype: kcoylenet