Print

---

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

Am 24.08.2015 um 22:21 schrieb Karen Coyle:

> This seems to me to be an implementation issue, and it would probably 
be
> preferable if BF allowed for more than one implementation choice. Ther
e
> is no reason why some implementations should not include these triples
> in their graphs:
> 
> bf:contributor [
>   a bf:Contributor ;
>   bf:role<http://id.loc.gov/vocabulary/relators/ill>  ;
>   bf:agent<http://id.loc.gov/authorities/names/1234>  ].

This superficially looks like a rather direct transformation of
MARC 700 with $4 and $0 (and nothing more is needed in a
"linked" situation). So what about

bf:contributor [
  a bf:Contributor ;
  bf:role<http://id.loc.gov/vocabulary/relators/ill> ;
  bf:role<http://id.loc.gov/vocabulary/relators/aut> ;
  bf:agent<http://id.loc.gov/authorities/names/1234> ;
  bf:agent<http://d-nb.info/gnd/9876> ].

(MARC semantics: two different roles for one bibliographic
identity accounted for in different authority files)


> <http://id.loc.gov/vocabulary/relators/ill>
>   skos:prefLabel "[email protected]" ;
>   a<http://www.loc.gov/mads/rdf/v1#Topic/>  .
> 
> <http://id.loc.gov/authorities/names/1234>
>   a<http://www.loc.gov/mads/rdf/v1#Authority/>  ;
>   skos:prefLabel "John Smith" .

and

<http://d-nb.info/gnd/9876>
  a<http://www.loc.gov/mads/rdf/v1#Authority/>  ;
  skos:prefLabel "Smith, John" .

Now an application generally may prefer to utilize information from
LC Authorities.

Membership information is part of LC Authority records

<http://id.loc.gov/authorities/names/1234>
  <http://www.loc.gov/mads/rdf/v1#isMemberOfMADSCollection>
<http://id.loc.gov/authorities/names/collection_LCNAF> .


and could be incorporated here, or would have to be provided
"externally" in the GND case:

<http://d-nb.info/gnd/9876>
  a<http://www.loc.gov/mads/rdf/v1#Authority/> ;
  bf:memberOfAuthoritativeCollection <http://d-nb.info/gnd> ;
  skos:prefLabel "Smith, John" .


So the trick built into LC Authorities seems to be not to
state that authority records belong to some authority file,
but to organize the resources described by the authority
records into collections.

VIAF on the other hand does take a different approach
(note presence or absence of the trailing slash in the URI):

<http://viaf.org/viaf/3456/> ;
 a <http://xmlns.com/foaf/0.1/Document> ;
 a <http://www.w3.org/2006/gen/ont#InformationResource> ;
 void:inDataset <http://viaf.org/viaf/data> ;
 foaf:primaryTopic <http://viaf.org/viaf/64049336"> .

and

<http://viaf.org/viaf/3456>
 a <http://schema.org/Person> ;
 dcterms:identifier "3456" ;
 schema:name "John Smith" ;
 ...
 schema:sameAs <http://id.loc.gov/authorities/names/1234> ;
 ... .

(and it also gives an alternative representation of the LC
Authority record:)

<http://viaf.org/viaf/sourceID/LC%7Cn++1234#skos:Concept>
  a <http://www.w3.org/2004/02/skos/core#Concept"> ;
  skos:inScheme <http://viaf.org/authorityScheme/LC>;
  skos:prefLabel "Smith, John, 1874-1951";
  foaf:focus <http://viaf.org/viaf/3456> .


So authority records can (and sometimes do) state
a) the authority control environment they belong to,
and they usually contain the strings needed
b) for use as label (heading, access point, ...)
c) as well as the identifier(s) assigned in that environment.

Bibframe could define three properties (or recommend the use
of specific general properties like rdfs:label) just to
normalize the situation.

Defining labels and identifiers as properties of the (non information)
resource makes sense in the context of a given authority control
system, but has to be dealt with some care in applications where
"authority control" from different systems exists in parallel:
The entities are to be identified, however their meta information
has to be kept in distinct compartments.

The Bibframe identifier approach intends to tackle this: I would
say that assigning resources to controlled collections (LC NAF),
assigning resource descriptions to a dataset (VIAF) or
assigning identifiers to identifier schemes (Bibframe proposal)
are technically different but semantically quite equivalent and
therefore legitimate approaches to inject "context" into our
descriptions. And I can perceive string identifiers and labels
assigned within a given controlled environment as "identifying
data" (as opposed to "universal" properties of the resource)

The question is, can the approaches be mixed? I.e. some
application can extract anything it needs from

bf:contributor [
...
 bf:agent<http://id.loc.gov/authorities/names/1234> ;
...].

but another one would need things spelled out explicitly by
added statements like:

<http://id.loc.gov/authorities/names/1234>
  a bf:Identification ;
  bf:xxxScheme bf:lcnafScheme ;
  bf:identifier "1234" ;
  bf:label "John Smith" .

(and perhaps adding
  rdfs:definedBy <http://id.loc.gov/authorities/names/1234>
  or
  schema:sameAs <http://id.loc.gov/authorities/names/1234>
like VIAF does to indicate that there is a information resource
determining the semantics / fixing the identity of that concept)

This would turn the concept <http://id.loc.gov/authorities/names/1234>
(a bf:Agent) into a bf:Identification at the same time: Would that
be something "usual" like trying to keep up a distinction between
informational resources and concept URIs in situations where the
"targets" don't cooperate (or operate in ways not known to the casual
user: "using authority data" in a Bibframe environment should be
possible without prior exams in the fine print of Semantic Web
standards...)?
Or would this rather mean (different equivalence relations depending
on which class is applied) a straight path to modeler's hell?

viele Gruesse
Thomas Berger

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

iJwEAQECAAYFAlXcJ0YACgkQYhMlmJ6W47PbeQQAs6oYyhcz2fdl0Oyh10DQTNQy
uZlN9y1DNGIUmK8XgcjlLAPV3MQse9ZplZvRB7CsnwtiOnNPrQYxeqNOhCuS4Igi
V1Ohkwm2k3aL4bbxg2fjpTtaEhnglNrv720b6SOVSVLqi8EMciKXtNSN5kQg0KHD
AnJ6z8cBxz2AlHAdhLo=
=+v5l
-----END PGP SIGNATURE-----