KNOWLEDGE ORGANIZATION: bringing order to the information universe

1
KNOWLEDGE ORGANIZATIONbringing order to the
information universe
2
FRAN ALEXANDER

• Library Assistant
• Reference Editor
• Editorial Director at Keesings
• MRes at UCL (DIS UCLIC)
• Taxonomy Manager at BBC

All views expressed here are entirely my own
personal views and in no way represent the BBC or
official BBC policy.
3
STRUCTURE OF THE AFTERNOON

• purpose of Knowledge Organisation (KO)
• some formal KO methods
• group exercise sorting out objects
• systematic structures
• break
• how do we go about making KO structures?
• complex subjects and citation ordering
• group exercise creating a classification
• representation and labelling

4
KNOWLEDGE ORGANISATION IS NOTHING NEW

• Hittite and Sumerian catalogues and lists
• hypertext links in The Talmud
• libraries in ancient Greece and Rome
• renaissance quests for Memory Palaces and
  Universal Languages
• Linnaean taxonomy
• Otlets Mundaneum and search engine service
• Ranganathan to Rosenfeld
• Internet of Things AI and onwards

5
(No Transcript)
6
WHY ORGANIZE?

• have you ever tried to find files on someone
  elses desktop?
• even if you only have a few hundred files,
  finding them again can take ages
• media archives have millions of files
• footage/recordings/documents that cant be found
  have no value
• free text search only takes you so far (scissors
  and scalpels)

7
WHY NOT JUST USE GOOGLE?

• synonyms and misspellings
• disambiguation - which Titanic? Budget and Spain?
• imperfect prior knowledge - Trying to learn about
  a topic
• aboutness - meaning beyond the words
• comprehensiveness
• audio-visual assets
• 2 billion users of the Internet

8
WHAT IS THIS FOOTAGE OF?

• Nothing will ever surpass the first flight I
  made on 6th October 1941. Dressed for the first
  time in overalls, helmet and goggles, I sat in
  the rear seat and bumped across the grass until
  the aircraft suddenly stopped bumping and we had
  left the ground behind. 35 minutes of ecstasy
  until we touched down.

9
PROFESSIONAL KNOWLEDGE ORGANIZATION

• a core function of the information professional
• to avoid chaos!
• how many published items?
• USA (2008) 288,000, UK (2009) 133,000 (up from a
  total in USA and UK in 2005 of 378,000)
• how many websites?
• 156 million (2008) 266 million (December 2010)
• to present resources in an orderly and
  predictable manner
• to enable access to specific content
• to aid retrieval of specific items
• to support exchange of information through the
  use of standard formats

10
HOW DO USERS LOOK FOR INFORMATION?

• Retrieval function of KO
• users may search for specific items - known item
  retrieval
• they may search for items characterized by some
  particular feature
• books by a certain author, document forms, etc.
• they may look for specific information
• Browsing function of KO
• they may want to see what is available
• they may not know what terms to use

11
HOW DOES KNOWLEDGE ORGANIZATION SUPPORT THESE TWO
APPROACHES?

• the processes of enabling access to knowledge
• labelling resources
• classification
• indexing
• tagging
• building vocabularies
• creating formal records to represent resources
• cataloguing
• bibliographic description
• metadata schemes
• creating systematic structures to hold
  information
• classifications, taxonomies, concept and topic
  maps, ontologies

12
LABELLING RESOURCES

• adding information to a resource about its
  subject content
• classification
• classification schemes and codes
• subject cataloguing
• subject heading lists
• indexing
• controlled vocabularies, thesauri, keyword lists
• metadata schema
• tagging
• usually uncontrolled

13
Group exercise categorising objects
14
SYSTEMATIC STRUCTURES FOR THE ORDERING OF
KNOWLEDGE

• sometimes there is a need to present information
  in a structured way
• physical organization materials in a physical
  collection
• listing presentation of items such as a subject
  bibliography or index
• display browsing interface of a digital
  collection

15
(No Transcript)
16
CREATING FORMAL RECORDS TO REPRESENT ITEMS

• listing characteristics of an item that represent
  it
• what its called? name, title
• who created it? author, creator
• who published it? publisher (commercial,
  institutional, personal)
• when and where? place of publication, web
  address
• whats it about? subject descriptors,
  classification codes
• physical attributes? size, dimensions, file
  type, references, illustrations
• representing these as fields in a database or
  equivalent structure
• using rules to ensure conformity of entries

17
(No Transcript)
18
(No Transcript)
19
(No Transcript)
20
(No Transcript)
21
SYSTEMATIC STRUCTURES FOR THE ORDERING OF
KNOWLEDGE

• it will be necessary to group items according to
  subject
• this is often described as classification or
  categorization
• the structure can be linear (as in a
  classification)
• the structure can be two-dimensional (as in a
  concept map)
• hypertext can be used to represent different
  levels of a hierarchy (as in taxonomies)

22
(No Transcript)
23
(No Transcript)
24
(No Transcript)
25
HOW DO WE MAKE A KO STRUCTURE?

• dont muddle the design of the interface with the
  structure of the information
• data must be well structured to support browsing
  and retrieval
• sequence of topics must be logical
• relationships between topics must be clear
• structure must be understandable and predictable

26
TOP-DOWN AND BOTTOM-UP CLASSIFICATIONS

• traditionally classifications were made by
  repeated subdivision of classes into smaller and
  smaller units
• this tends to create rather rigid and abstract
  classifications
• modern methods tend to work by clustering or
  grouping concepts to form classes
• this method creates more flexible systems, more
  closely related to reality

27
TREE STRUCTURES
28
CLUSTERING STRUCTURES
29
(No Transcript)
30
SORTING AND GROUPING

• this is the first stage in organizing a
  collection of objects or concepts
• different attributes may be used as the basis of
  the classification
• a whole variety of different (but quite valid)
  classifications can be made by varying the
  criteria for arrangement
• this explains why classifications almost always
  show cultural bias of some kind

31
Can you name the odd one out?

• trumpet
• violin
• French horn
• trombone

32

• trumpet
• violin
• French horn
• trombone

33

• London
• China
• Brazil
• France

34

• London
• China
• Brazil
• France

35

• French
• Spanish
• Hebrew
• English

36

• French
• Spanish
• Hebrew
• English

37
ENTITY CLASSIFICATIONS

• used in scientific classifications
• used to arrange objects or entities themselves
• each entity has a unique place in the
  classification
• typical examples are
• chemical elements
• minerals
• astronomical bodies
• biological organisms

38
(No Transcript)
39
(No Transcript)
40
ASPECT CLASSIFICATIONS

• bibliographic or documentary classifications
  arrange first by subject field
• objects or entities are scattered
• a rabbit could appear in many different places in
  a library classification
• zoology - cookery agriculture - fashionpet
  keeping - conjuringmedical research -
  mythology
• the rabbit is a distributed relative

41
(No Transcript)
42
(No Transcript)
43
(No Transcript)
44
(No Transcript)
45
Break
46
COMPLEXITY IN SUBJECT MATTER

• life would be easier if subjects were simple
• the subjects of most documents or resources are
  very complicated
• decisions must be made about the location of a
  complex topic
• this involves giving priority to some attributes
  over others
• this should be done
• consistently
• with the needs of the users in mind
• there are lots of examples of this problem in the
  real world, not just in the information world

47

• dog
• cow
• potato
• rabbit
• Michelangelo
• Titian
• Dante
• Van Gogh

48

• dog
• cow
• potato
• rabbit
• Michelangelo
• Titian
• Dante
• Van Gogh

49

• dog
• cow
• potato
• rabbit
• Michelangelo
• Titian
• Dante
• Van Gogh

50
(No Transcript)
51
CITATION ORDER

• the order in which the various constituent
  elements of a subject are listed when arranging,
  classifying, or indexing documents
• citation order should be consistent and
  predictable in order to retrieve things
• in a taxonomy, the choice of citation order will
  determine which are the top levels of the taxonomy

52
(No Transcript)
53
Examples of citation order in physical
arrangement

• Tesco arrangement is by storage /preservation
• fresh food
• tinned food
• chilled food
• frozen food
• W. H. Smith arrangement is by form
• newspapers
• magazines
• books
• videos
• DVDs/CD-ROMs

54
Group exercise

• Creating classifications card sorting technique

55
DOCUMENT REPRESENTATION

• indexing terms or descriptors
• informal tagging
• a formal record of the document or resource,
  e.g. a bibliographic record, metadata schemes, a
  database structure with separate fields for the
  different elements

56
FORMAL SYSTEMS OF REPRESENTATION

• the last century has seen the development of a
  number of standard formats for records, and, more
  recently, for metadata
• these esnable interoperability and the exchange
  of information
• were concerned here primarily with the structure
  of content, and not the exchange formats per se

57
INTERNATIONAL AGREEMENTS

• most countries had their own standards for
  records
• in 1950 UNESCO held an international conference
  aimed at achieving a universal record of
  information output
• also aimed to standardize bibliographic document
  formats
• Anglo-American Cataloguing Rules and the MARC
  format
• potential uniformity at a global level

58
Catalogue record showing MARC21 tags
59
MARC tags indicate fields in foreign language
record
60
MARC tags indicate fields in foreign language
record
61
(No Transcript)
62
METADATA STANDARDS

• in the late 20th century similar initiatives were
  put in place for metadata for digital resources
• best known is the Dublin Core, developed by OCLC
  in Dublin Ohio
• DC close in format and coverage to the MARC
  standard
• many metadata standards have been created to
  cover other aspects of digital resources
• Semantic Web and Linked Data are underpinned by
  metadata standards

63
METADATA UNIVERSE
64
Metadata embedded in coding for website
65
SUBJECTIVE AND OBJECTIVE METADATA

• none of these standards make any provision for
  the specification of subject
• this is mainly because subject cannot be inferred
  automatically from the resource itself
• subject evaluation is a very subjective process
  and presents some specific problems
• a separate subject indexing tool must be used
• this usually takes the form of a controlled
  vocabulary

66
SUMMARY

• knowledge organization matters
• as digital content explodes, the need for KO
  increases
• natural language usually too fuzzy for
  effective retrieval
• many systems and methods of controlling
  vocabulary used, various strengths and
  weaknesses, many viewpoints

67
Image credits

• Labels http//www.flickr.com/photos/mrsmagic/5870
  198525/
• Electrolux Design Lab - The robotic kitchen
  assistant http//www.flickr.com/photos/electrolux
  -design-lab/3811899536/sizes/l/in/photostream/
• Hieroglyphs at the Karnak temple
  http//www.flickr.com/photos/tamburix/2900907735/
• Metadata Universe http//www.dlib.indiana.edu/je
  nlrile/metadatamap/