RDA and the semantic Web
Lectio magistralis in Library Science
by Gordon Dunsire
Florence University, Florence, Italy
4th March, 2014
In the beginning …
3?
Toronto, 1997
General Material Designator (US)
General Material Designator (UK)
Content vs Carrier / Haworth
Logical analysis / Delsey
Chief source of
information
Class of
materials
ISBD(G)
Series
Subseries Item
Type of publication
Structure
$£$£$£!!!
? of AACR3
Functional Requirements for
Bibliographic Records
User
tasks
Work
Expression
Manifestation
Item
Content attributes
Carrier attributes
JSC, 2005
AACR3
Item
Semantic Web (Berners-Lee)
Structured
collections
of information
Sets of inference rules
Automated reasoning
Web of linked data
Web of linked documents
Web of linked computing devices
Resource Description Framework (RDF)
“This work has author Gordon Dunsire”
Triple!
Subject – Predicate - Object
? ex:1Bd38
Subject
“This work”
Predicate
Object
“has author”
? ex:ZX3G5
“Gordon Dunsire” ? “literal”
Uniform Resource Identifier (URI)
For machines
For humans
ex:“is derivative
work of”
ex:
“This work”
ex:
“This work”
ex:
“That work”
ex:“has derivative work”
ex:“has title”
“RDA and the
Semantic Web”
ex:“has author”
“Gordon Dunsire”
One
ex:“has name”
giant
global ex:
“G. J. Dunsire”
graphGordon
ex:“has author”
Dunsire
ex:“is author of”
ex:“has
country of birth”
ex:“has alternate
name”
ex:
Scotland
Data vs Ontology
ex:
“This work”
ex:“has author”
ex:
Gordon
Dunsire
Data triple
“has author”
Ontology
triple
URI
ex:
hasAuthor
ex:“has label”
ex:“has type”
“Property”
RDA vocabulary example
rdact:
1010
ex:“has label”
ex:“has label”
ex:
“is related to”
“notated music”@en
“schriftlich fixierte Musik”@de
rdact:
1016
ex:“has label”
“tactile notated music”@en
Inference rules: sub-property
ex:
Property1
ex:
“is subproperty
of”
ex:“has label”
ex:
Property2
ex:
“This work”
“has author”@en
ex:“has label”
ex:“has author”
“has creator”@en
“Gordon Dunsire”
ex:“has creator”
ex:
“Gordon Dunsire”
“This work”
RDF vocabularies: DCT + RDFS
Dublin Core Terms + RDF Schema
rdfs:label
dcterms:
creator
“Creator”
rdfs:range
Source of URIs
rdfs:label
“Agent”
dcterms:
Agent
“A resource that
acts or has the
rdfs:comment power to act.”
Inference rules: range
dcterms:
creator
rdfs:range
ex:
“This work”
ex:
Gordon
Dunsire
dcterms:
agent
dcterms:creator
rdfs:type
ex:
Gordon
Dunsire
dcterms:
Agent
Data model meeting
Panizzi
British Library, London, 30 April – 1 May 2007
Members of RDA and Semantic Web communities
“… agreed that RDA and DCMI should work
together …”
(Dublin Core Metadata Initiative)
Benefits expected
The library community gets a metadata standard that
is compatible with the Web Architecture and that is
fully interoperable with other Semantic Web
initiatives
The DCMI community gets a libraries application
profile firmly based on the DCAM and FRBR (which
will be a high profile exemplar for others to follow)
The Semantic Web community get a significant pool
of well thought-out metadata terms to re-use
There is wider uptake of RDA
Activities
Development of an RDA Element Vocabulary
Development of an RDA DC Application Profile
based on FRBR and FRAD
Disclosure of RDA Value Vocabularies using
RDF/RDFS/SKOS
Opaque URI
(only for machines)
Compact URI = rdact:1010
…
DCAM and RDA
Dublin Core Abstract Model
<indecs>
Basic RDA metadata structures
ex:
ExpressionURI
rdae:
“has content
type”
rdact:
1010
rdfs:label
skos:inScheme
“notated music”@en
rdaterms:
RDAContentType
High-level metadata structures
Unconstrained
rdau:
“has title”
Constrained by FRBR (domain)
rdfs:
“domain”
rdac:
“Manifestation”
rdfs:
“domain”
sP
rdam:
“has title”
rdau:
“has title proper”
sP
rdam:
“has title proper”
sP
sP = rdfs:”sub-property of”
Relationship designators
rdau:
“has creator”
sP
rdaw:
“has creator”
sP
rdac: rdfs:
rdfs: rdac:
“Work” “domain”
“range” “Agent”
rdaw:
“has designer”
rdau:
“has designer”
sP
sP = rdfs:”sub-property of”
Link-ability
Entity
Domains
Properties
Relationships
Attributes
Range No Range
Work
Expression
235
235
200
190
35
45
Manifestation
210
50
160
Item
Agent
50
225
40
175
10
50
Triple clusters
Triple chains
Covering all media
Literals
rdaw:
“has related work”
Rich detail
rdaw:
“has accompanying work relationship with”
rdaw:
“is complemented by (work)”
rdaw:
“is screenplay (work)”
rdaw:
“is video screenplay (work)”
sub-property ladder
Interoperability
rdau:
“has creator”
rdaw:
“has creator”
rdau:
“has performer”
rdau:
“has actor”
rdau:
“has voice actor”
???
dc:
“creator”
dcterms:
“creator”
To do
Publish more RDA value vocabularies
Develop cross-entity elements
E.g. Work-Expression relationship designators
Develop Application Profile for RDA “core”
Develop RDF elements for aggregated
statements (RDA, ISBD, MARC, …)
E.g. Place of publication -> Publication statement
Achievements: 5+ years on
Rich set of elements and value vocabularies for
“bibliographic control”
At global, “universal” level
“a metadata standard that is compatible with the
Web Architecture and that is fully interoperable
with other Semantic Web initiatives”
“a significant pool of well thought-out metadata
terms to re-use”
Wider uptake of RDA? …
Thank you!
gordon@gordondunsire.com