Towards Interoperability

1
Towards Interoperability

• Paul Miller
• Collections ManagerArchaeology Data ServiceArts
  Humanities Data Service (UK)
• collections_at_ads.ahds.ac.uk

2
Arts Humanities Data Service

• AHDS is funded by the UK Higher Education Funding
  Councils, and comprises
• Archaeology Data Service (York et al.)
• History Data Service (Essex Data Archive)
• Oxford Text Archive (Oxford)
• Performing Arts Data Service (Glasgow)
• Visual Arts Data Service (Farnham)
• an Executive (Kings College, London).

3
Aims of the AHDS

• AHDS is
• a distributed collection of disciplinespecific
  services
• each with additional responsibility servicewide
  for a data type
• a model for decentralised data archiving and
  access
• AHDS is building
• a single gateway to Arts Humanities data of
  interest to UK academics
• data remain distributed in many locations, linked
  by means of ISO 23950 Z39.50, Dublin Core, etc.

4
Data in the Arts Humanities (1)

• Arts Humanities data encompass a wide range of
  types and formats, including
• text
• raw, SGML markedup, PDF, etc
• databases
• flat file, relational, spatial, temporal, GIS,
  etc
• images
• manuscripts, works of art, remote sensing, film,
  video, etc
• sound
• recordings, MIDI, etc.

5
Data in the Arts Humanities (2)

• These data not only span diverse technical
  formats, they are also
• constructed within differing conceptual
  frameworks
• geographies, theoretical paradigms, etc
• Creator may not be quite synonymous with
  Author
• recorded following different and inconsistent
  cataloguing practices
• described using many different metadata
  systems, if formally described at all.

6
Data in the Arts Humanities (3)

• These data are too diverse to be effectively
  retrieved by means of any one search system
• but
• a description of the core metadata for each
  resource may prove comparable within and between
  disciplines, facilitating effective resource
  discovery.

7
What is Metadata?

• meaningless jargon, or
• a fashionable term for what weve always done,
  or
• a means of turning data into information, and
• data about data, and
• the name of a film director (Luc Besson), and
• the title of a book (The Lord of the Flies)
• etc
• Metadata means many things to many people
• AHDS is currently concerned primarily with
  resource discovery metadata.

8
The Dublin Core (1)

• probably the best tool for providing core
  resource discovery metadata
• international, crossdomain effort to achieve
  definition of a core element set
• defines 15 core elements
• allows optional qualification of these through
  addition of thesauri and lookup tables (SCHEME),
  subclassification of the elements (SUBELEMENT)
  and metadata language (LANG)
• hopes to capture the essence of any resource
• but is it too Core?

9
The Dublin Core (2)

• Title
• Creator
• Subject
• Description
• Publisher
• Contributors
• Date
• Type

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

http//purl.org/metadata/dublin_core/
10
Models of Implementation (1)
DC.title DC.creator DC.subject DC...
A User
A Resource
Dublin Corerecord
11
Models of Implementation (1)
DC.title DC.creator DC.subject DC...
1
A User
A Resource
Dublin Corerecord
2
12
Models of Implementation (2)
DC.title DC.creator DC.subject DC...
A User
A Resource
Dublin Corelayer
mapping/ crosswalk
13
Models of Implementation (2)
DC.title DC.creator DC.subject DC...
1
A User
A Resource
Dublin Corelayer
2
mapping/ crosswalk
14
Models of Implementation (3)
DC.title DC.creator DC.subject DC...
1
A User
A Resource
Dublin Corefilter
mapping/ crosswalk
15
Models of Implementation (3)
DC.title DC.creator DC.subject DC...
1
A User
A Resource
Dublin Corefilter
2
mapping/ crosswalk
16
AHDS/UKOLN Workshops (1)

• one of few attempts to discover what users and
  depositors require from a Core Element set
• created jointly by AHDS and UKOLN to
• resolve AHDS particular and immediate
  problems
• explore the wider issues of crossdomain,
  interdisciplinary, distributed resource
  discovery.
• Dublin Core used as reference set, but
• participants examined both where it failed to
  meet their needs and where it offered more than
  required
• DC was not seen as a replacement for other
  standards
• participants were offered the opportunity to add
  new elements, rather than cram existing ones
  inappropriately

17
AHDS/UKOLN Workshops (2)

• Six workshops held
• two (digital sound and moving images) for PADS,
  one for each of the other Service Providers
• integrated with ongoing technical deliberations
• Invitees included
• experts in holding and describing domainspecific
  data
• those depositing these data
• current and potential users of the data
• me.

18
AHDS/UKOLN Workshops (3)

• Draft reports widely circulated for comment
• Final reports from each workshop now available
• http//ahds.ac.uk/
• http//www.ukoln.ac.uk/metadata/
• Integrated report recently published
• Discovering Online Resources Across the
  Humanities a practical implementation of the
  Dublin Core. Edited by Paul Miller Daniel
  Greenstein.

19
Assessing the Dublin Core (1)

• Dublin Core is not
• a replacement for existing detailed metadata
  schemes
• they still have an (important) role to play
• a means for describing data sets, concepts, or
  subject issues in great detail
• the answer to all our problems (!)
• Many of the problems encountered by workshops
  were not with Dublin Core itself, but were
  related to more generic data description and
  cataloguing issues
• In many cases, workshops began by confusing these
  external issues with those integral to Dublin
  Core.

20
Assessing the Dublin Core (2)

• Dublin Core is
• a useful means by which discrete data types and
  sets may be described in a comparable fashion
• small enough to remain manageable, yet extensible
  enough to (hopefully) be suitably descriptive
• a fascinating example of interdisciplinary and
  international cooperation
• (if used in conjunction with the concepts of the
  Warwick Framework) an extremely powerful means of
  drawing complex metadata and data together,
  facilitating access and reuse.

21
Assessing the Elements (1)

• Dublin Core found to be fit for purpose
• definitions found to be unsatisfactory
• interpreted too differently by the six workshops
• AHDS interpretation of the definitions in our
  report
• will be fed back in to wider Dublin Core
  community
• CREATOR and CONTRIBUTORS found to be confusing
• notions of primary intellectual responsibility
  difficult to assign
• some workshops suggest a single element, NAMES,
  instead. In the short term, AHDS will ignore
  CONTRIBUTORS.

22
Assessing the Elements (2)

• SUBJECT open to abuse
• easily overloaded with many terms from many word
  lists
• potential conflict with COVERAGE and TYPE
• AHDS suggests distinction between quantifiable
  and nonquantifiable aspects of space and time
• what is the subject of Hamlet, anyway?!
• PUBLISHER means different things to different
  people
• AHDS to compile and suggest controlled list of
  publishers.

23
Assessing the Elements (3)

• DATE not sufficient for requirements
• creation of original work? publication date of
  version later digitised? release date of
  electronic version? update cycle dates?
• AHDS to compile and suggest controlled list, in
  collaboration with DC date working group
• TYPE confusing collection of concepts.

24
Assessing the Elements (4)

• FORMAT concept extended to nondigital
• AHDS suggests inclusion of film running times,
  video formats, etc where absolutely required
• SOURCE and RELATION need clarified
• AHDS Service Providers hold different notions of
  source
• both could be misused with overinclusion of
  useful relationships
• one suggestion is that transcription SOURCE,
  whilst derivation RELATION

25
Assessing the Elements (5)

• COVERAGE is complex
• close involvement with DC Coverage working group
• is the European Parliament building the SUBJECT
  or COVERAGE of a photograph?
• what are the usefully recorded spatial COVERAGEs
  for a Frankish bowl made in Aachen, excavated in
  Trier and on view in the British Museum?
• The Holy Roman Empire? Aachen? France?
  Germany? Trier? British Museum? London?
  Europe?
• what is The Holy Roman Empire?
• temporal COVERAGE ?

26
Assessing the Elements (6)

• RIGHTS essential
• AHDS is developing a simple rights management
  coding scheme to be used in conjunction with a
  mandatory link to detailed rights management
  information for each individual resource.

27
Assessing the Qualifiers

• optional extensibility of SCHEME and SUBELEMENT
  found to be essential. LANG useful in certain
  cases
• every use of a SCHEME or SUBELEMENT increases
  Dublin Cores value to one discipline, and
  reduces interoperability with the others
• many SCHEMEs and SUBELEMENTs identified in
  workshop reports
• integrated report attempts to aggregate these,
  moving back towards interoperable generalisations
• where is middle ground between value to one
  discipline and the overreaching goal of
  interoperability?

28
Moving Forward (1)

• Resources of interest to AHDS are
• diverse
• an archaeological excavation database and a
  recording of the Berlin Philharmonic playing Ode
  to Europe
• distributed
• a database physically mounted in York, the
  Scottish NMR in Edinburgh, and the Shetland
  Amenities Trust SMR in Lerwick all accessible to
  the user in Pisa or Antwerp
• living
• a Local Authority SMR, updated every day
• rarely in HTML
• so Harvesting is not the best solution.

29
Moving Forward (2)

• Z39.50 seen as the solution
• preserves distributed nature of resources
• capable of expressing many data types
• (relatively) large body of implementation
  experience
• DC1 profile may solve current
  compartmentalisation problems to some extent
• allows easy integration with CIMI, Aquarelle,
  etc
• having gained sufficient expertise, targets may
  be implemented at collaborating organisations,
  extending system functionality.
• probably ISite (its free, and spatially aware)

30
A Model
AHDSGateway
HDS
OTA
PADS
VADS
ADS
31
A Model
WWW browser
AHDSGateway
HDS
OTA
PADS
VADS
ADS
(Z Target)
32
A Model (extended for ADS)
SCRAN (Scotland)Museums (worldwide)HSYS
(England)ADAP (USA)NGDF (UK) NUTS/ SABE
(EU)Thesauri CoE, GII etc.plus local ADS
collections
Z Target /WWW browser
AHDSGateway
etc.
ADSGateway
HDS
OTA
PADS
VADS
CIMITestbeds
33
Moving Dublin Core Forward (1)

• AHDS wishes to make a commitment to Dublin Core
• e.g. Miller Greenstein 1997
• Other European projects wish to use Dublin Core,
  too
• but they and AHDS often need to convince
  partner organisations to participate
• this is difficult with Dublin Core perceived as
  impermanent, and subject to change every six
  months with each new workshop...

34
Moving Dublin Core Forward (2)

• Dublin Core needs (now!!!)
• single recognised, workable, implementation to
  which we can all refer
• AHDS offers Miller Greenstein as one
  possibility upon which we might build
• clear authority(s) for resolving conflict
• AHDS recommends a small oversight group as final
  arbiters of community consensusbuilding efforts
• defined procedures by which change may be made to
  a stable DC Version 1
• even if only for Europe...

35
Moving Dublin Core Forward (3)

• So
• shall we stop talking
• and just
• DO IT?!