An Introduction to Dublin Core

1
An Introduction to Dublin Core Making Sense of
Metadata, Society of Archivists EAD/Data
Exchange SIG London, Thursday 17 November
2005 Pete Johnston Research Officer, UKOLN,
University of Bath
UKOLN is supported by
www.bath.ac.uk
2
An Introduction to Dublin Core

• A brief history
• What is Dublin Core, really?
• The DCMI Abstract Model
• Encoding Dublin Core metadata
• DC Application Profiles
• DC in practice

3
A Brief History
4
A brief history (1)

• Mid 1990s rapid growth of World Wide Web
• Challenge of resource discovery
• search engines providing many hits, but little
  precision
• recognition that library approach to cataloguing
  could not scale to Web resources
• 1995 OCLC/NCSA Workshop in Dublin, Ohio
• interdisciplinary consensus on 13 "metadata
  elements"
• for discovery of "document-like objects"
• relatively simple, usable by non-cataloguers
• 1996 OCLC/CNI Workshop in Dublin, Ohio
• expand to 15 elements
• explicitly cross-domain
• for discovery of broad range of "resources"

5
The Dublin Core Metadata Element Set

• Title
• Subject
• Description
• Creator
• Publisher
• Contributor
• Date

• Type
• Format
• Identifier
• Source
• Language
• Relation
• Coverage
• Rights

6
A brief history (2)

• 1997-2000 Development of notion of
  "qualification"
• tension between simplicity and complexity
• element refinement
• Narrow the meaning of a DC element
• e.g. "date modified" v "date"
• encoding scheme
• Provide additional information about a value
• e.g. that a subject is a Library of Congress
  Subject Heading
• the "Dumb-Down" principle
• Rules for transforming "qualified" description
  into "simple" description
• the "One-to-One" rule
• A DC description describes exactly one resource

7
A brief history (3)

• 1997-2000 What is a "resource"?
• e.g. Can the DCMES be applied to people?
• DCMI Type Vocabulary
• Collection, Dataset, Event, Image (Still or
  Moving), Interactive Resource, Service, Software,
  Sound, Text, Physical Object
• But still fairly non-prescriptive
• 1998- Emergence of Resource Description Framework
  (RDF)
• 2000-2001 "Grammatical Principles" as informal
  data model

8
A brief history (4)

• 2000-2005 Development of notion of DC
  "Application Profile"
• tailoring metadata standards for context
• providing local guidelines, constraints
• combining components from different sources
• 2003-2005 Formalisation of DCMI Abstract Model
• concepts used in DC metadata
• different types of terms used in DC metadata
• how those terms used in combination to construct
  descriptions

9
What is Dublin Core, really?
10
Dublin Core is...

• a conceptual framework/set of rules...
• DCMI Abstract Model
• describes how to use certain types of terms
• ... to make statements...
• ... that form descriptions (of resources)
• a "core" vocabulary/set of terms...
• managed by DCMI (Usage Board)
• growing (relatively) slowly as new requirements
  arise
• each identified by a Uniform Resource Identifier
  (URI)
• a set of specifications for representing or
  encoding DC metadata descriptions in various
  formats

11
DCMI Abstract Model(a slightly simplified view)
12
DCMI Abstract Model

• A description
• describes exactly one resource
• may specify a resource URI
• consists of a set of statements

13
DCMI Abstract Model Descriptions
14
DCMI Abstract Model

• A statement must contain
• a reference to a property
• property URI
• all DC "elements" are properties
• properties may be defined by agencies other than
  DCMI
• a reference to a second resource (value)
• value URI, and/or
• one or more value representations
• value string
• rich representation

15
DCMI Abstract Model Statements
16
DCMI Abstract Model

• A statement may contain
• a reference to a vocabulary encoding scheme
• vocabulary encoding scheme URI
• type of value
• a reference to a syntax encoding scheme
• syntax encoding scheme URI
• how value string is interpreted

17
DCMI Abstract Model Statements
18
DCMI Abstract Model

• A description describes one resource
• Applications typically based on description sets
• groups of descriptions
• where the described resources may be related in
  some way
• Description sets encoded or serialised as records
• according to rules of binding

19
(No Transcript)
20
Encoding Dublin Core metadata(a very brief
introduction!)
21
DCMI Abstract Model and Bindings

• For transfer between applications, descriptions
  must be represented as digital objects
• Binding maps between constructs in conceptual
  model and components in a digital format
• Two way
• encoding application description set -gt record
• decoding application record -gt description set
• DCMI currently provides three "encoding
  guidelines" specifications
• Other agencies may also provide bindings

22
Using X/HTML meta link elements

• The set of meta/link elements represent a single
  DC description.
• The resource described is the X/HTML document in
  which the metadata is embedded.
• Each meta/link element represents a single
  statement
• Property and Encoding Scheme URIs encoded as
  prefixed names

ltlink rel"schema.DC" href"http//purl.org/dc/ele
ments/1.1/" /gtltlink rel"schema.DCTERMS"
href"http//purl.org/dc/terms/" /gt ltmeta
name"DC.title" content"A guide to DC metadata"
/gt ltmeta name"DCTERMS.audience"
content"information managers" /gt ltmeta
name"DC.language" scheme"DCTERMS.ISO639-2"
content"eng" /gt ltlink rel"DCTERMS.references"hr
ef"http//dublincore.org/documents/dcq-html" /gt
23
Using the DC-XML format

• Supports only limited subset of Abstract Model
  (revision forthcoming)
• The container element, here ltmetagt, represents a
  single DC description.
• Each child element represents a single statement
• Property URIs and Encoding Scheme URIs encoded as
  XML QNames

lt?xml version"1.0"?gtltmeta xmlns"http//www.ukol
n.ac.uk/metadata/dcdot/" xmlnsxsi"http//w
ww.w3.org/2001/XMLSchema-instance"
xmlnsdc"http//purl.org/dc/elements/1.1/"gt
ltdcidentifiergthttp//example.org/doc/1234/lt/dcid
entifiergt ltdctitlegtA Guide to DC
Metadatalt/dctitlegt ltdclanguage
xsitype"dctermsISO639-2"gtenglt/dclanguagegt
ltdctermsreferencesgthttp//dublincore.org/document
s/dcq-htmllt/dctermsreferencesgt lt/metagt
24
Using the Resource Description Framework (RDF)

• Specifications for DC in RDF do exist
• but currently work in progress to
• resolve ambiguities
• revise in light of DCAM

25
Dublin Core Application Profiles
26
DC Application Profile

• Implementers adapt metadata standards to the
  context of their application
• Tension between localisation and interoperability
• A DC Application Profile
• specifies the terms (properties,
  vocabulary/syntax encoding schemes) used in a
  class of description sets
• describes how those terms are used
• supplementary information on how properties
  applied/interpreted in context
• constraints on occurrence of properties
• constraints on values and value representations
  (encoding schemes)

27
DC Application Profiles Examples

• "Simple Dublin Core"
• use of the 15 properties of the DCMES
• all optional and repeatable
• values represented by value strings
• no vocabulary or syntax encoding schemes
• UK eGMS
• use of selected properties from DCMI
  vocabularies, additional properties
• guidelines on use of properties
• some properties mandated/recommended
• some vocabulary encoding schemes
  mandated/recommended
• guidance on content of value strings

28
DC Application Profiles Examples

• JISC Information Environment Service Registry
  (IESR) Metadata Schema
• supports description of several related resources
  (Collection, Service, Agents)
• use of selected properties from DCMI
  vocabularies, selected properties from RSLP CD
  vocabularies, some properties created for IESR
• for each subject resource type, guidelines on use
  of properties
• some properties mandated/recommended
• many vocabulary encoding schemes
  mandated/recommended

29
(No Transcript)
30
DC in Practice
31
Dublin Core in X/HTML

• Initial implementation focused on DC-in-HTML
• Robot crawls individual HTML pages to extract
  metadata
• But today little/no use by large Web search
  engines
• Problems of spamming/trust
• Lack of take-up by authors/publishers
• Success of full-text crawling/indexing, esp.
  Google!
• However, some use in controlled domains
• Intranets
• Trusted groups of providers (e.g. eGMS)
• Embedding DC in XHTML useful if you know a search
  engine exploits it

32
(No Transcript)
33

• Picture Australia- images "related to all things
  Australian" from 40 cultural agencies"
• central search service based (initially at
  least) on crawling HTML-embedded DC metadata
• providers migrating to OAI-PMH
• currently hybrid approach?

http//www.pictureaustralia.org/
34
(No Transcript)
35
(No Transcript)
36
Dublin Core and OAI-PMH

• Open Archives Initiative Protocol for Metadata
  Harvesting (OAI-PMH)
• Fairly simple mechanism for sharing metadata
  records between applications
• Has origins in e-prints community
• Built on HTTP, XML
• Allows a harvester to ask a repository for all or
  some of its metadata records (in a specified
  metadata format)
• i.e. supports "incremental harvesting"
• "Give me all your records updated since
  yyyy-mm-dd"
• "OAI-DC" (Simple DC) is mandatory format
• But no limitation on format that can be
  transferred (as long as can be described by XML
  Schema)

37
Repositories
38

• OAIster (University of Michigan)
• "academically-oriented digital resources"
• "5,947,627 records from 557 institutions"
  (2005-11-15)

http//oaister.umdl.umich.edu/
39
(No Transcript)
40
(No Transcript)
41
Summary

• DCMES/"Simple DC" as a "core" for discovery of
  wide range of resources
• "Simple DC" is, by definition, simple!
• Limitations in terms of functions/services that
  can be offered
• DCMI Abstract Model provides a framework for
  extensibility and modularity
• A DC Application Profile describes a real-world
  usage of that model

42
An Introduction to Dublin Core Making Sense of
Metadata, Society of Archivists EAD/Data
Exchange SIG London, Thursday 17 November
2005 Pete Johnston Research Officer, UKOLN,
University of Bath
UKOLN is supported by
www.bath.ac.uk