Identifying the Identifiers

1
Identifying the Identifiers

• Douglas CampbellNational Library of New Zealand
  Te Puna Matauranga o Aoteaora
• DC-2007, 29th August 2007

255685
2
Designing Identifiers

• Identifier Theory
• Identifier Qualities
• Identifier Design Checklist

3
Identifiers 101

• Identify this

Chair ? Seat Plastic and metal
thing 1 metre tall with four legs
4
Identifiers 101

• Identify this

Chair Seat Plastic and metal thing 1 metre tall
with four legs The one on the left The blue one
5
Identifiers 101

• Identify this

Chair Seat Plastic and metal thing 1 metre tall
with four legs The one on the left The blue
one The stacking chair Vitra Tom Vac Chair Asset
no. 1123-33
6
(No Transcript)
7
Identifiers 999
Identify these
Application Profile (namespace)
Metadata schema (namespace)
Age (data field)
Economics (social networking tag)
Donor (relationship)
Economics (subject authority)
Boy with dog (photograph in collection)
Romeo and Juliet (FRBR work)
Boy with dog (digitised photo)
Craft Society (harvested, archived website)
Picasso, Pablo (1881-1973) (name authority)
Film director (agent role)
8
Identifiers in Communication

• Identifiers are part of communicating to refer
  to a thing

The goal is samenessLatin idem (the same)
facere (to make)
9
Prerequisites for identifying

• To Identify we need to Differentiate
• To Differentiate we need to Compare sameness
• To Compare we need to Define / Describe
• To Describe we need to- Observe
  characteristics eg. size, location- Interpret
  characteristics eg. smell, topic- Assign new
  characteristics eg. name, logo, id string

10
Identify by comparing characteristics

• We identify by comparing the sameness, or not, of
  characteristics

?
Bring a seat in

• Can sit on

?
?
11
Identify by comparing characteristics

• We identify by comparing the sameness, or not, of
  characteristics

?
Bring an office seat in

• Can sit on
• Office

?
?
12
Promoting characteristics to identify

• Use description characteristic(s) as surrogate
  (substitute)
• A symbol to represent the thing
• Convenience compare single identifiers
  instead of lots of characteristics
• Promotion changes the characteristics role
  purpose is differentiation, not description

13
Context
?
Invitation Street Party at No. 4
?
?
?
14
Describing and Identifying
Description
characteristics
characteristics
describe
contexts
associate
15
Identifier Definitions

• Identifier
• A stated association between a symbol and a
  thing that the symbol may be used to
  unambiguously refer to the thing within a given
  context.
• Thing Any entity, idea, action, resource,
  object, etc.
• Symbol Any mark, token, sensory stimulus,
  character string, etc.
• Identifier system
• Policies, processes, and/or mechnisms for
  assigning, managing, and using identifiers.

16
Semiotics

• Study of how we communicate using signs and
  symbols
• We use symbols with no intrinsic meaning,we add
  meaning around these symbols

17
Semiotic Triangle

• Nothing is a sign unless it is interpretted as a
  sign - Charles Pierce

Symbol
Object
18
Semiotic Triangle
agent
Concept
Symbol
Object
19
Semiotic Triangle
agent
Concept
Symbol
Object
(implied relationship)

• An identifier is a thought

20
Semiotic Triangle
remembrance of association and context
agent
Concept
Symbol
Object
(implied relationship)

• Identifiers are the manifestation of the act of
  identifying

21
The Deconstructed Identifier

• The desconstructed identifier has six aspects
• A Thing
• A Symbol built from characteristics in a
  description
• An Associaton between the symbol and the thing
• A Context
• An Agent that states the association and context
• A Remembrance of the association and context

22
Identifier Qualities

• Scope
• Uniqueness
• Granularity
• Intelligence
• Actionability
• Persistence
• Extensibility
• Context

23
Scope

• Draw identifier from description, but be clear
  what is being described
• Newspaper article

?
DB
24
Multiple scopes in one record

• A MARC record contains multiple scopes(thats
  not wrong, just be aware of it)
• Dublin Cores one-to-one rule is useful
  alternative

100 Creator 245 Title 650 Subject 856 URL
Vocabulary term
25
Uniqueness

• Often want to refer to a thing unambiguously

Thing
Thing
Thing
26
The unique Johns

• Cannot uniquely identify John

John
John
John
John
27
Uniqueness

• A thing has only one identifer
• An identifer only relates to one thing

Thing
Thing
Thing
28
Uniqueness

• A thing has only one identifer
• An identifer only relates to one thing

Thing
Thing
Thing
29
Global uniqueness

• Wrap naming authority identifier around local
  identifiers

Org 1123
Org 2123
Org 3123
30
Granularity

• Question How deeply should we break groups into
  separately identified things?
• Answer If you have an need to identify it, then
  identify it!!
• Methodology examples FRBR, ltindecsgt
• In practice, the identifier system may dictate
  constraints

• Journal
• Newspaper
• Page
• Article
• Photo

31
Intelligence

• Adding roles of description and remembrance
  to identifiers
• Intelligent Dumb Semantic Transparent
  Opaque
• Intelligent identifiers are time-dependent
  based on your world view at the time, eg. country
  names, gay, email address

32
Location as Identifier

• Lazy identifiers, e.g.
• System file path to HTML file Web URL
• Location on shelf identifier
• Location identifiers are needed to access the
  thing,but may not be the best identifier to
  publicise
• Dilution different identifiers for copies of
  the same thing

a.com/x.html b.org/y.html c.com/z.jsp
33
Sidebar http URIs

• URI Universal Resource Identifier
• URL Universal Resource Locator (a kind of
  URI)
• Many original URLs were lazy identifiers (using
  file location)
• Many are now more considered identifiers they
  just happen to start with http// called
  http URIs
• Often are intelligent identifiers, so beware (as
  discussed)

34
Actionability
Live Dead ActionableResolvableDe-referenceabl
e
?
35
Persistence

• How long does an identifier need to live?
• How do we keep it alive that long?
• Not a technology issue, is a commitment issue
• Need policies for handling changes in environment
• When an identifier is retired
• When the thing itself changes
• When the identifier system becomes obsolete
• When the custodian of the identifier changes
• Degree of mutability (allow identifiers to be
  re-associated?)

36
Extensibility

• Persistence of identifier systems
• Risk from unanticipated demand
• Risk from re-use in unanticipated situations
• Risk from changes to the environment
• Future-proof by including capacity to be adapted
• As generic form as possible
• Hooks for community-defined extensions
• Consider scalability
• Follow international standards
• Keep application independent

37
Context

• Remembrance of association and its context
• Remembrance of context alongside or combined
• .
• except
  need to know what a URN is!
• Often context is missing, assuming the reader can
  infer it!
• Multiple identifiers in multiple contexts is not
  undesirable
• Sameness is different for different communities
• Though helpful for similar contexts to be
  combined

38
Checklist for designing identifiers

• Audience
• Consider how the identifiers are intended to be
  used and potential downstream uses
• Scope
• Determine the thing(s) being identified/described
  (scope, granularity)
• Context
• Determine the context(s) things are being
  identified within (granularity). For example,
  is it a concept/item/component/instance/etc., or
  what communities will it serve?

39
Checklist for designing identifiers

• Overlap
• Consider the relationship of the identifiers to
  other similar identifiers and/or contexts,
  consider merging
• Persistence
• Determine the expected identifier lifespan and
  strategies to preserve the relationship to the
  associated thing for that long (e.g. commitment
  level, resourcing, and policies)
• Design the identifier system
• Identifier structure design uniqueness,
  intelligence, actionability, persistence,
  extensibility, and communication of context
• Addressability - combine identifiers or
  standalone identifiers?
• Support - policies, processes, and mechanisms

40
Checklist for designing identifiers

• Assign locally
• implementation (within your scope of control)
• Global uniqueness
• Wrap local identifiers with global authority
  identifiers for wider use
• Use them
• i.e. avoid using equivalent identifiers that may
  cause duplication or confusion!

41
Conclusion
42
Describing and Identifying
Description
characteristics
characteristics
describe
contexts
associate
43
The Deconstructed Identifier

• The desconstructed identifier has six aspects
• A Thing
• A Symbol
• An Associaton
• A Context
• An Agent
• A Remembrance

remembrance of association and context
agent
Concept
Symbol
Object
(implied relationship)
44
Action List

• Look backwards
• Identifier audit
• Look forwards
• Identifier goals requirements
• Take action!
• Make identifiers unique internally

45
DCMI Identifiers Community

• DCMI not the right place to do identifiers
  workBut is a good place to disseminate
• Key issues raised in DC Conference Special
  Session
• Identification vs. resolution
• Importance of being able to resolve
• New DCMI Identifiers Community October 2007
• http//dublincore.org/groups/identifiers/
• http//dublincore.org/identifierswiki/

46
Identifiers Special Session

• Thursday 30th August 2007, 11.30am 1pm,
  Karimata Room
• Identifier work
• NISO identifiers roundtable John Kunze
• DOIs URNs Juha Hakala
• Identifier principles Stu Weibel
• Identifiers at National Library of New Zealand
  Douglas Campbell
• Discussion on identifier issues
• What are the issues or areas of confusion?
• What are the areas we DO understand?
• What are some best practices?

47
Questions?

• http//www.dcmipubs.org/ojs/index.php/pubs/article
  /view/34
• douglas.campbell_at_natlib.govt.nz