Is that a Reference Model in your Pocket

1
Is that a Reference Model in your Pocket?

• The CIDOC CRM the IFLA FRBR
• Tony Gill

12 May 2003
2
Outline

• The problem of cultural information diversity
• Data models, ontologies and reference models
• CIDOC CRM overview
• IFLA FRBR overview
• Comparative analysis
• Real-world envisioned applications

3
Cultural infodiversity

• Cultural information held by museums, libraries
  and archives is necessarily heterogeneous
• Curatorial approaches
• Subject disciplines
• Granularity
• Level of detail
• Data structure
• Data content values

4
Cultural infodiversity

• Cultural information held by museums, libraries
  and archives is necessarily heterogeneous
• Curatorial approaches
• Subject disciplines
• Granularity
• Level of detail
• Data structure
• Data content values
• Infodiversity is good!

5
Cultural infodiversity

• No single (meta)data schema fits all
• Significant conceptual overlaps
• e.g. People, places, events, objects,
  relationships
• Traditional compromise approach
• Generic simple descriptions for initial
  discoverye.g. Dublin Core
• Rich domain-specific descriptions for depthe.g.
  EAD, MARC
• Access by lowest common denominator
• Crosswalk proliferation ? mapping madness!

6
Data models

• Data models are structures for information
• Good data models mirror the reality the data
  are attempting to describe
• Different data modeling methodologies
• Entity-Relation (Relational, SQL, RDBMS)
• Object-Oriented (O-O, OODBMS)
• Semantic Networks (RDF, DAMLOIL)
• Few museum, library or archive information
  standards are based on robust data models!

7
Ontologies

• A branch of metaphysics concerned with the
  nature and relations of being
• Merriam Webster Collegiate Dictionary
• Ontology co-opted by knowledge representation
  computer science communities
• A specification of a conceptualization
• Tom Gruber, A translation approach to portable
  ontologies Knowledge Acquisition, 1993
• Thesauri classification schemes are ontologies!

8
Reference models

• Generalized data models or ontologies
• Emphasis on high-level conceptual analysis of a
  domain
• Intended for reference more than implementation
• Help disambiguate dialogue and identify common
  concepts across heterogeneous
• Data repositories
• Descriptive schema
• Data models
• Communities
• Validate existing new schema, data models etc.

9
CIDOC Conceptual Reference Model

• Object-oriented domain ontology
• Formalizes the semantics required to describe
  objects and relationships in the museum/cultural
  heritage (as defined by ICOM) context
• NOT a metadata standard! But can be used to
  express metadata standards
• Represents over a decade of development
• Based on ICOM/CIDOC International Guidelines for
  Museum Object Information The CIDOC Information
  Categories
• Scope covers rich information exchange between
  museums, libraries and archives

10

• The primary role of the CRM is to serve as a
  basis for mediation of cultural heritage
  information and thereby provide the semantic
  'glue' needed to transform today's disparate,
  localised information sources into a coherent and
  valuable global resource.
• Martin Doerr Nick Croftshttp//cidoc.ics.forth.
  gr/

11
CRM overview

• CRM v3.4 comprises 81 Classes interlinked by 130
  Properties
• Classes inherit properties from their parents, or
  Superclasses
• Event-centric and empirical observations about
  the world
• Short-cuts, for typically incomplete knowledge
• Highly extensible through Sub-typing of classes
  and properties
• E.g. Jane Hunter, Combining the CIDOC CRM MPEG
  7 to Describe Multimedia in Museums, MW2002
  Proceedings
• Ideally suited to RDF implementation

12
Scope of the CRM

• Intended scope Exchange and integration of
  scientific documentation about museum collections
• Scientific means sufficient depth precision
  for research
• Museum defined by ICOM
• Includes contextual information
• Includes exchange between museums, libraries
  archives
• Excludes administrative information, e.g. visitor
  statistics
• Practical scope The set of extant data sets and
  structures used in museum documentation
• The curated knowledge of museums

13
Types
refine
Actors
Conceptual Objects
Physical Entities
Appellations
identify/name
participate in
affect
Temporal Entities
location
within
occur at
Places
14
(No Transcript)
15
(No Transcript)
16
Mapping to the CRM

• Mappings entail deconstruction of original
  records
• Artifact-centric nature of descriptions discarded
• Implicit entities (especially events) made
  explicit
• Mappings to existing standards
• EAD
• IFLA FRBR
• SPECTRUM
• AMICO
• And others

17
Making implicit concepts explicit

• The element DC.Creator implies
• An Actor, who created something
• An Actor Appellation by which to identify the
  creator
• An Event, the act of creation
• Some Man-Made Stuff, the physical or conceptual
  thing that was created and is being described by
  the DC record
• E24 Physical Man-Made Stuff p108 was produced by
  E12 Production Event p14 carried out by E39 Actor
  p131 is identified by E82 Actor Appellation
• E28 Conceptual Object p94 was created by E65
  Creation Event p14 carried out by E39 Actor p131
  is identified by E82 Actor Appellation

18
E52 Time-Span
E53 Place
February 1945
7012124
E39 Actor
took place at
within
E38 Image
participated in
E7 Activity
E39 Actor
refers to
Crimea Conference
falls within
E65 Conc. Creation
E39 Actor

carried out
within
has created
E31 Document
E52 Time-Span
Yalta Agreement
1945-02-11
19
Benefits of the CRM

• Elegant and simple compared to comparable
  Entity-Relation model
• Coherently integrates information at varying
  degrees of detail
• Readily extensible through object-oriented class
  typing and specializations
• Richer semantic content allows (some) inferences
  to be made from fuzzy data
• Designed for semantically lossless mediation of
  heterogeneous cultural heritage information

20
CRM learning curve

• Model necessarily complex in order to model the
  broad domain of cultural heritage information
• Object-oriented modeling paradigm unfamiliar
  compared to entity-relation modeling
• Just similar enough to be confusing!
• Object-oriented models can be implemented using
  relational DBMS
• Notation problems
• Difficult to express mappings textually

21
Introducing FRBR

• FRBR Functional Requirements for Bibliographic
  Records
• IFLA International Federation of Library
  Associations
• 7 year study final report published 1997
• Comprehensive review of requirements for
  bibliographic description and control motivated
  by
• Development of automated library systems
  networks
• Evolving user needs
• Economic pressures
• Conceptual framework expressed as an
  entity-relationship (E-R) model
• Doesnt address requirements for authority
  records

22
And its cousin, FRANAR

• Functional Requirements Numbering of Authority
  Records
• FRANAR study underway results not yet published
• Eventual goal merge FRBR FRANAR into a single,
  comprehensive model for bibliographic description

23
The FRBR entities

• Group 1 The products of intellectual or artistic
  endeavour
• Work
• Expression
• Manifestation
• Item

• Group 2 Those responsible for production or
  custodianship of group 1 entities
• Person
• Corporate Body
• Group 3 Subjects of works
• Concept
• Object
• Event
• Place

24
FRBR group 1 entities primary relationships
Work
is realized through
Expression
is embodied in
Manifestation
is exemplified by
Item
25
FRBR group 1 entities primary relationships
Work
is realized through
Expression
is embodied in
Manifestation
abstract, intangible
is exemplified by
physical, tangible
Item
26
FRBR group 2 entities responsibility
relationships
Group 2
Work
is created by
Person
Expression
is realized by
CorporateBody
Manifestation
is produced by
Item
is owned by
27
FRBR entities subject relationships
Work
Has as subject
28
Comparison between CRM FRBR

• Mapping comparison between CIDOC CRM IFLA
  FRBR
• Patrick Lebuf, The Book, the Bug the Bangle
  A parallel a paradox CIDOC CRM Sharing the
  Knowledge Symposium
• FRBR incomplete, CRM (aims to be) comprehensive
• FRBR confuses entities with their names
• CIDOC CRM has explicit Appellation class
  hierarchy
• FRANAR to rectify

29
Comparison between CRM FRBR

• CRM exhibits museum focus on specific objects and
  their contexts FRBR has library emphasis on the
  properties of manifestations of works
• Bibliographic description closer to the taxonomic
  discourse of natural history museums than the
  artifact/context focus of art cultural history
  museums!

30
Real-world applications

• Conceptual reference
• Disambiguating dialogue (especially between
  domain technology experts)
• Validation of schema (c.f. Patrick LeBouef, FRBR)
• Information exchange
• Canonical master mappings
• Expression in XML or RDFS (CIDOC CRM)

31
Real-world applications

• System schema design CIDOC CRM
• RLG Cultural Materials
• Finnish National Gallery Database
• City of Geneva MusInfo Project
• Germanische Nationalmuseum Nuremberg
• Monument Inventory Data Standard
• Heritage Data Dictionary
• CLIO Cultural Documentation System,
  ICS-FORTH/Benaki Museum
• System schema design IFLA FRBR
• RLG RedLightGreen
• ARTstor The Illustrated Bartsch
• AustLit Gateway

32
Envisioned applications

• Mediation systems agents
• Meaningful queries and results across distributed
  heterogeneous data sources
• Semantic web for culture?
• Gradual transition from record-centric
  documentation to knowledge networks
• How to navigate potentially unbounded networks?
• How to maintain links within potentially
  unbounded networks?