The Registry Data Model

1
The Registry Data Model

• Pete Johnston, UKOLNp.johnston_at_ukoln.ac.uk

2
The registry data model and its expression in RDF

• The registry data model
• Resources to be described
• Relationships between resources
• Attributes/properties of those resources
• RDF RDFS
• Expressing the model in RDF
• The Registry vocabulary

3
Assumptions

• Model based on a Dublin Core-like view of a
  metadata record
• A metadata record is a collection of pairs of
  attributes (or properties) and values
• Title Julius Caesar Author William
  Shakespeare
• The value of a property may be associated with an
  encoding scheme that enables a more accurate
  interpretation
• Subject Drama in English. Shakespeare, William
  Texts (SchemeLCSH)
• Date 2002-03-05 (SchemeW3CDTF)

4
Assumptions

• Each property is an occurrence of a metadata
  element
• Title, Author, Subject
• Metadata elements are defined/managed as members
  of metadata element sets created for a purpose
• Semantic relationships may exist between metadata
  elements (equivalence, refinement etc.)
• Implementers optimise their use of metadata
  elements by
• Drawing metadata elements from multiple element
  sets
• Constraining usage of metadata elements in
  context

5
The registry data model

• Builds on earlier work of DESIRE, SCHEMAS
  projects, ongoing DCMI interest, MEG registry
  project
• A simplification of complexity, a basis for
  comparison
• Types of resource described
• Agencies
• Element Sets
• Elements
• Encoding Schemes
• Values in controlled vocabulary Encoding Schemes
• Application Profiles
• Element Usages

6
The registry data model

• The registry reads, indexes and exposes
  descriptions of instances of these classes of
  resource
• i.e. metadata about individual elements sets,
  elements etc
• That metadata is stored in Schemas
• A Schema will (probably) provide metadata about
  multiple resourcese.g. the Elements in an
  Element Set
• The metadata about a resource may be spread
  across multiple Schemas (though not using the
  registry Client)
• Metadata about Schemas not (presently)
  processed/exposed by registry

7
Agency
Element Set
Encoding Scheme
Element
Value
Element Usage
AppProfile
8
Key relationships

• An Element is-Element-Of exactly one Element Set
• An Element Usage is-Usage-In exactly one
  Application Profile
• An Element Usage uses exactly one Element
• The several Element Usages in one Application
  Profile may use Elements from different Element
  Sets.

9
Agency
An organisation or individual responsible for
managing one or more Element Sets, Application
Profiles or Encoding Schemes
10
Agency relationships

• Element Set --- is-Managed-By (m-1) ? Agency
• Encoding Scheme --- is-Managed-By (m-1) ? Agency
• Application Profile --- is-Managed-By (m-1) ?
  Agency

11
Agency attributes
12
Agency example
13
Element Set
A set of metadata Elements that is managed as a
coherent unit by an Agency. The Elements of an
Element Set are functionally related, by virtue
of having been defined for the purpose of
usefully describing the characteristics of a
resource
14
Element Set relationships

• Element Set --- is-Managed-By (m-1) ? Agency
• Element --- is-Element-Of (m-1) ? Element Set

15
Element Set attributes
16
Element Set attributes
17
Element Set example
18
Element
A formally defined term that is used to describe
a characteristic or attribute of a resource
19
Element relationships

• Element --- is-Element-Of (m-1) ? Element Set
• Element --- associated-Encoding-Scheme (m-m) ?
  Encoding Scheme
• Element --- refines (m-1) ? Element
• Element Usage uses (m-1) ? Element

20
Element attributes
21
Element attributes
22
Element example
23
Encoding Scheme

• A set of contextual information or parsing rules
  that aids in the interpretation of the value of a
  metadata Element.
• Encoding Schemes include
• formal notations or parsing rules, which define
  precisely how a lexical representation of a value
  is to be interpreted, and
• controlled vocabularies, which enumerate a list
  of values

24
Encoding Scheme relationships

• Encoding Scheme --- is-Managed-By Agency (m-1) ?
  Agency
• Element --- associated-Encoding-Scheme (m-m) ?
  Encoding Scheme
• Element Usage --- associated-Encoding-Scheme
  (m-m) ? Encoding Scheme
• Value type (m-1) ? Encoding Scheme

25
Encoding Scheme attributes
26
Encoding Scheme attributes
27
Encoding Scheme example
28
Controlled Vocabulary Value
An individual value or term in a controlled
vocabulary

• Value type (m-1) ? Encoding Scheme

29
Controlled Vocabulary Value attributes
30
Controlled Vocabulary Value example
31
Application Profile
A set of Element Usages that is managed as a
coherent unit by an Agency. An Application
Profile is optimised for the resource description
requirements of a particular application or
context. Like the Elements of an Element Set,
the Element Usages within an Application Profile
are functionally related, by virtue of having
been defined for the purpose of usefully
describing a resource. Within an Application
Profile, the Element Usages may reference
Elements from multiple Element Sets
32
Application Profile relationships

• Application Profile --- is-Managed-By Agency
  (m-1) ? Agency
• Element Usage --- is-Usage-In (m-1) ? Application
  Profile

33
Application Profile attributes
34
Application Profile attributes
35
Application Profile example
36
Element Usage

• A deployment of a (previously defined) metadata
  Element in the context of a particular domain or
  application. The used Element may be tailored for
  the context by
• a narrowing of its semantic definition
• association with specified datatypes or Encoding
  Schemes
• specification of obligation/occurrence constraints

37
Element Usage relationships

• Element Usage --- is-Usage-In (m-1) ? Application
  Profile
• Element Usage --- uses (m-1) ? Element
• Element Usage --- associated-Encoding-Scheme
  (m-m) ? Encoding Scheme

38
Element Usage attributes
39
Element Usage attributes
40
Element Usage example
41
Annotation
42
Statements in RDF are global

• DCMI says
• The DC metadata element http//purl.org/dc/element
  s/1.1/title has a definition A name given to
  the resource.

http//purl.org/dc/elements/1.1/title/
A name given to the resource
43
Statements in RDF are global

• RSLP says
• (In our context of use) the DC metadata element
  http//purl.org/dc/elements/1.1/title has a
  definition The name of the collection.

http//purl.org/dc/elements/1.1/title/
The name of the collection
44
Statements in RDF are global

• My learning object schema says
• (In our context of use) the DC metadata element
  http//purl.org/dc/elements/1.1/title has a
  definition A name given to this learning object.

http//purl.org/dc/elements/1.1/title/
A name given to this learning object
45
Statements in RDF are global

• Merging three statements.

http//purl.org/dc/elements/1.1/title/
46
RDF Vocabulary Description Language (RDF Schema)

• RDF Vocabulary Description Language provides
• Classes properties to describe classes,
  properties
• Does not provide
• Class for a vocabulary
• Property for membership of a vocabulary
• rdfsisDefinedBy used to indicate a resource
  defining the subject resource. This property may
  be used to indicate an RDF vocabulary in which a
  resource is described. (emphasis added)
• Class for a schema

47
Expressing the registry model in RDF

• Use classes and properties from existing
  vocabularies where possible
• Define application-specific classes and
  properties where required
• rdfsClass for each of principal entities
• regElementSet, regAppProfile, regElementUsage,
  regEncodingScheme, regAnnotation
• rdfProperty for each attribute and relation

48
Expressing the registry model in RDF

• Each Encoding Scheme is itself an rdfsClass
• Each Controlled Vocabulary Value is an instance
  of that class
• i.e. a specific Controlled Vocabulary Value is
  related to an Encoding Scheme by an rdftype arc

49
Expressing the registry model in RDF

• Overlapping of datatyping and Encoding Scheme?
• RDF Core Work on literal datatyping
• Temporary workaround of application-specific
  properties
• regdatatype
• regassociatedEncodingScheme

50
Element Sets Elements
http//example.org/dcmi/
http//example.org/dcmi/elementset/1
http//purl.org/dc/elements/1.1/title
51
Application Profiles Element Usages
52
Application Profiles Element Usages
53
Merging Application Profiles Element Sets
54
RDF, RDFS and the CORES Registry

• RDF provides a way of representing statements
  about resources as nodes and arcs in graphs
• RDF/XML provides a syntax for recording RDF
  graphs as XML so that they can be exchanged
  between applications
• RDF Vocabulary Description Language (RDF Schema)
  provides mechanisms for describing properties and
  the classes of resources they describe
• The CORES-MEG Vocabulary provides additional
  properties and classes specifically for
  describing resources of the types identified in
  the data model

55
RDF, RDFS and the CORES Registry

• The CORES Client enables an author to
• create and edit RDF descriptions using these
  vocabularies
• serialize (output) those descriptions as RDF/XML
  and read serialized descriptions into the
  authoring environment
• submit descriptions to the registry server (as
  RDF/XML)
• The CORES Registry
• reads, stores and indexes RDF data
• pushed to server by CORES Client
• pulled by server from RDF/XML available on the
  Web
• (potentially) pushed by other applications using
  the publication API
• provides query interfaces to that data for human
  readers and software tools

56
The Registry Data Model

• Pete Johnston, UKOLNp.johnston_at_ukoln.ac.uk