2012-01-05. Frozen archive - links may not resolve - see directory of files at MoinMoin wiki archive

> AlistairMiles/AnalysisNotes20081011

Notes on RDA work, 2008-10-11.

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

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

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):

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...

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