Task: To revise the RDF expression of the cataloguer scenarios http://dublincore.org/dcmirdataskgroup/Scenarios using a FRBR vocabulary in conjunction with the RDA vocabulary.
Looking at http://metadataregistry.org/schema/list.html I find a "FRBR Entities" schema http://metadataregistry.org/schema/show/id/5.html but no declared classes or properties.
Looking at the sandbox http://sandbox.metadataregistry.org/ schemas I find nothing related to FRBR, in vocabularies I find
-
[1] FRBR Entities http://sandbox.metadataregistry.org/vocabulary/show/id/49.html
-
[2] FRBR Relationships http://sandbox.metadataregistry.org/vocabulary/show/id/90.html
-
[3] FRBR Relationships (as concepts) http://sandbox.metadataregistry.org/vocabulary/show/id/64.html
-
[4] FRBR User Tasks http://sandbox.metadataregistry.org/vocabulary/show/id/69.html
To express cataloguer scenarios, I need FRBR entities and relationships expressed as RDFS classes and properties (i.e. a schema).
From Ian Davis et al I find
-
[5] Expression of Core FRBR Concepts in RDF http://vocab.org/frbr/core
-
[6] Expression of Extended FRBR Concepts in RDF http://vocab.org/frbr/extended
Probably worthwhile doind a comparison between [1]+[2] (Gordon's FRBR, interpreted as classes and properties) and [5] (IanD's FRBR), to compare naming conventions and to see if any properties have been missed.
On initial impression, IanD's FRBR [5] looks very clean, nice naming conventions and at a glance looks very complete in terms of semantics.
Comparison of FRBR Schemas by Gordon and Ian Davis
Began comparison, published at http://spreadsheets.google.com/pub?key=pWXJnkpaKdGk4kKVbATNgGg
Classes
Both have classes for group 1 entities: Work, Expression, Manifestation, Item.
Aside, consult http://dublincore.org/dcmirdataskgroup/analysisTask1 for information on FRBR.
Both have classes for group 2 entities: Person, CorporateBody.
Both have classes for group 3 entities: Concept, Event, Object, Place
In addition, Ian's vocab has a class Endeavour which is the superclass of each of the group 1 classes, and is also equivalent to the union of the group 1 classes. N.B. in Karen's analysis http://dublincore.org/dcmirdataskgroup/analysisTask1 this looks like it corresponds to BibliographicResource.
In addition, Ian's vocab has a class ResponsibleEntity, which is the superclass of CorporateBody, and is also equivalent to the union of CorporateBody and Person; I expected ResponsibleEntity to also be a superclass of Person, maybe this an oversight on Ian's part. I.e. it makes sense for something like ResponsibleEntity to be the superclass of the group 2 classes. N.B. in Karen's analysis http://dublincore.org/dcmirdataskgroup/analysisTask1 this is called Agent.
In addition, Ian's vocab has a class Subject, which is the superclass of each of the group 3 classes, and is also equivalent to the union of the group 3 classes. This also appears in Karen's analysis http://dublincore.org/dcmirdataskgroup/analysisTask1 .
N.B. Ian's vocab also includes lots of connections to other vocabularies, e.g. frbr:Concept is a subclass of skos:Concept; frbr:Person is equivalent to foaf:Person.
Suggestions (I guess to IFLA):
-
that a FRBR schema be declared with the following classes: Work, Expression, Manifestation, Item, CorporateBody, Person, Concept, Object, Place, Event.
-
that the schema also declare a class which is the superclass of the group 1 classes, named either Endeavour or BibliographicResource.
-
that the schema also declare a class which is the superclass of the group 2 classes, named either ResponsibleEntity or Agent.
-
that the schema also declare a class which is the superclass of the group 3 classes, named Subject.
-
I suggest following a minimum commitment approach to semantics, which means *not* declaring any of the equivalent class assertions, at least for now
For now, I will work from the classes declared in Ian's FRBR core vocabulary [5].
Properties
Using Gordon's FRBR relationships vocabulary http://sandbox.metadataregistry.org/concept/list/vocabulary_id/90.html [2]...
Comparison, published at http://spreadsheets.google.com/pub?key=pWXJnkpaKdGk4kKVbATNgGg
Ian's vocab has properties for all FRBR relationships, i.e. complete coverage. For each FRBR relationship Ian has two properties (inverses), with the exception of frbr:subject (Ian has no frbr:subject).
In addition, Ian has three generalisation properties: frbr:relatedEndeavour, frbr:responsibleEntity and frbr:responsibleEntityOf.
Ian has applied a concise, consistent naming convention for properties. This makes coding the scenarios much simpler.
Gordon has hasAnimation and isAnimation of, which I can't find FRBR relationships for.
For the scenarios, I will use Ian's vocabulary.
I suggest IFLA use the same design principles and naming conventions as Ian has established, as these make the vocabulary easy to work with.
I.e. I suggest IFLA declare a FRBR schema with properties as named in Ian's FRBR Core schema. I also suggest IFLA state domain and range assertions as in Ian's FRBR Core.
N.B. Ian has also declared a FRBR extended vocabulary http://vocab.org/frbr/extended ... from a quick glance, looks like refinements of classes and properties given in FRBR core. For now, I won'y use any of Ian's FRBR extended vocabulary, only the FRBR core.
Cataloguer Scenario 1
Using http://metadataregistry.org/schemaprop/list/schema_id/1.html ...
Published updated RDF expression at Scenarios/1.
Issues...
-
Should rdarole:author be a sub-property of frbr:creator ?
-
What vocabulary to use for the form of work?
-
Found both originalLanguageOfTheWork and languageOfWork, appear identical function.
-
What vocabulary to use for language of work?
Found http://downlode.org/Code/RDF/ISO-639/ .. ? Old dublin core patterns for RFC1766 or ISO639-2, e.g. http://dublincore.org/documents/dcq-rdf-xml/? New dublin core patterns http://dublincore.org/documents/dc-rdf/ using dcterms:ISO639-2 as a datatype? Options at http://esw.w3.org/topic/Languages_as_RDF_Resources ? Given that LOC is registration authority for ISO 639-2, can we ask them to declare URIs for languages?
See also http://dublincore.org/documents/dcmi-terms/#ses-ISO639-2
-
What vocabulary to use for properties of a person? FOAF? VCARD? FRAD? ...
-
What's the difference between languageOfExpression and contentLanguage?
-
What vocabulary to use to describe properties of place? GEO...
-
Found rda:Publisher as property, expected lower case initial letter for property names.
-
What vocabulary to use to describe corporate bodies, e.g. publishers?
-
For extent, use extentOfText, with literal?
-
Found both identifierOfWork and identifierForTheWork.
-
No need for different properties for identifier for each type of entity, e.g. identifierForTheWork, identifierForTheManifestation etc., only need single identifier property.
-
What properties to use in authority record for a person?
-
Strange that rda:title says associated with manifestation.