On Tue, Dec 2, 2008 at 7:38 AM, Sebastian Rahtz
> Stuart A. Yeates wrote:
>> <place rdf:about="http://clas-lgpn2.classics.ox.ac.uk/placeid/d1e77">
>> <person rdf:about="http://clas-lgpn2.classics.ox.ac.uk/id/V-65834">
>> <start> -0425 <start>
>> <end> -0375 <end>
> if you're going to do that, why not stick with the original TEI?
Because unlike TEI, a TEI-like RDF vocabulary inferred (using OWL)
from CIDOC-CRM RDF would:
(a) be data-centric rather than document-centric (notions of sequence
and document structure are lost if not explicit)
(b) allow any tags anywhere (throw away the TEI schema)
(c) be a description-logic (well understood computationally)
(d) be amenable to the full range of RDF tools, databases and query
engines (think SPARQL)
(e) have the inter-operation issues with CIDOC-CRM ironed out up-front
(because it's syntatic sugar for the underlying CIDOC-CRM)
(f) allow arbitrary co-mingling with other RDF vocabularies (such as
dc:creator, mets:techMD, frbr:realizer, mods:Description, ...)
The lossy nature of (a) means that the TEI document is not recoverable
from TEI-like RDF, unless you take inordinate care to include both the
full text and header ordering / nesting information in the RDF, which
A TEI-like RDF vocabulary has the advantage (over raw CIDOC-CRM RDF)
that practitioners who're already working in TEI can use familiar
terms in RDF. They'll have slightly different semantics, but that
can't be avoided.
You're probably aware that the NZETC uses CIDOC-CRM topic maps (which
are like RDF) for essentially all extra-document metadata, including
document structure (for page-breaking decisions, navigation, etc),
named entities (people, places, organisations, ships, etc, all linked
against our authority database), filetypes (pdf, tei, etc) and
languages / character encodings (en, mi, latin, SAMPA, etc).
We are actively evaluating a move from topic maps to RDF.
I should also add that it's been suggested to me out-of-band that we
focus on getting he underlying representation right before focusing on
the syntatic sugar (as I have a tendency to).
In your original mapping Sebastian, both the sex and the
nymRef="#Zwa1lios" fell by the wayside. Was this debliberate?