Print

Print


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


Am 05.08.2014 20:18, schrieb Denenberg, Ray:

> Thomas Berger said:
> “…except that (1) and (3) must allow for multiple values,”  (string identifier and URI form)
> 
> Seems to me this is asking for trouble (matching the string with the URI).
> Why not instead provide multiple bf:IdentifierDescriptor resources (and declare them
> equivalent)?

In your example

<http://example.com/xyz/book1> a bf:Instance ;
  <http://example.com/xyz/identifiers/book1/identifier1>
     a bf:IdentifierDescriptor ;
     bf:identifierScheme <http://id.loc.gov/vocabulary/identifiers/isbn> ;
     bf:identifierValue "9780099483793" ;
     bf:uriFormOfIdentifier “urn:isbn:9780099483793”.

you provieded an arbitrary string version of a certain ISBN. And one
of many possible equivalent URIs by RFC-3187 (actually not, as since
Juha pointed out the urn:isbn scheme has not yet been extended to
ISBN-13).

And I provided some more equivalent forms for the identifier
manifest on your book:

<http://example.com/xyz/identifiers/book1/identifier1>
 bf:identifierValue "978-0-09-948379-3", "0-09-948379-3";
 bf:uriFormOfIdentifier "URN:ISBN:0-09-948379-3".


and of course there is "0099483793" and what we find imprinted in most
english language publications, namely "978 0 09 948379 3", and
"0 09 948379 3" (unfortunate example since the check digits of the
ISBN-10 and ISBN-13 forms coincide). And of course there are
representations prefixed by the string "ISBN ".

Most of us will probably agree that the dashed, blanked and undelimited
forms are just notational variants of each other. Additionally my point
of view is that also "978-0-09-948379-3" and "0-09-948379-3" are just
notational variants of /one/ ISBN (I have been referring to them as
"abstract identifiers" in previous messages, please compare this with
the distinction between "value space" and "lexical space" e.g. in the
W3C XML Schema Datatypes document (< http://www.w3.org/TR/xmlschema-2/ >)).

So for me the bf:IdentifierDescriptor as a resource would correspond
to a distinct element of the /value space/ of the identifier system
defined by the bf:identifierScheme and which element this is would
be indicated by the bf:identifierValue literal residing in the lexical
space for that identifier.

If we had an XSD ISBN datatype the typed literals "978-0-09-948379-3"
and "0-09-948379-3" would be the same anyway, so it would be nonsense
to state more than one of them. Using the appropriate datatype also
would obliviate the chore of explicitly naming the identifierScheme
and we then could completely dispense the bf:identifierDescriptor and
remain with


<http://example.com/xyz/book1> a bf:Instance ;
  bf:identifier "978-0-09-948379-3"^^xsd:ISBNtype .

Defining individual datatypes for any odd identifier system however
seems not feasible, therefore we are sticking to plain string literals
for the identifierValues and I see no harm to supply one bf:IdentifierDescriptor
with several bf:identifierValues, especially since we usually
denote (or transcribe) the ISBN-10 and its equivalent ISBN-13
within any description: Keeping equivalent ISBNs syntactically
together helps keeping non-equivalent ISBNs (for print and eBook and
the like) apart, I think.

In an earlier message I proposed to set up with each "registered"
identifierScheme an auxiliary bibframe address space for that
system, providing "common" lexical values for the identifiers.

so I could state
<http://example.com/xyz/identifiers/book1/identifier1> owl:sameAs
  <http://bibframe.org/registry/isbn/978-0-09-948379-3>

and this actionable URI could pull in

<http://bibframe.org/registry/isbn/978-0-09-948379-3>
     a bf:IdentifierDescriptor ;
     bf:identifierScheme <http://id.loc.gov/vocabulary/identifiers/isbn> ;
     bf:identifierValue "9780099483793", "0099483793", "978-0-09-948379-3",
"0-09-948379-3", "978 0 09 948379 3", "0 09 948379 3" ;
     bf:uriFormOfIdentifier “urn:isbn:9780099483793”, “URN:ISBN:0-09-948379-3”.

and so on: A script simply would expand the identifier extracted from
the URN and expand it to certain well known forms. (Of course the same
content would be delivered when resolving
<http://bibframe.org/registry/isbn/0-09-948379-3> or
<http://bibframe.org/registry/isbn/9780099483793>:
I'm not proposing some covert normalization of the permitted lexical
values by distinguishing some URIs over others.

Using that bibframe service infrastructure in a sense one could emulate
what the hypothetical XSD datatype could provide:

<http://example.com/xyz/book1> a bf:Instance ;
  bf:identifier <http://bibframe.org/registry/isbn/978-0-09-948379-3>;

(I just have to remember that for accessing the actual values I have either
to retrieve bf:identifierValues from the URI given or extract them
from the URI with non-RDF tools)

 %--

As for the equivalence of  bf:identifierValue and bf:uriFormOfIdentifier
I'm not certain: RFC-3187 seems to state (in the language of 2003) that
there is a transformation from ISBN value space to URI space and that
apart from URI equivalence rules there are equivalence rules specific
to URN:ISBN URIs. These are the common equivalences implemented by
ignoring dashes, but I'm not sure about blanks and prefixes like "ISBN ".
So it is probably best to think of independent sets of equivalence
rules for bf:identifierValue s and bf:uriFormOfIdentifier s for any
given scheme.

viele Gruesse
Thomas Berger
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iJwEAQECAAYFAlPhST8ACgkQYhMlmJ6W47ODlAP+MeQAjEv9A7qenHL24ePG31OP
JJ/O79gPsEqw1HyvSuxKC3ARk3pAZ1SGFRd9RAueXVJP6ZkmWc38oCZ2cEEfZNiS
zLq5DEKiXSOyzkfOCftsE4LsyllZnI8lAZ+xySsVFCP0jN9BvGzxPI0C23kVqj94
6G4slkSGyRFNqZy71I0=
=UK8b
-----END PGP SIGNATURE-----