Workshop 3 DAY 3

1
Workshop 3 DAY 3

• Philip Calvert
• School of Information Management
• Victoria University of Wellington
• New Zealand

2
What is a Thesaurus?

• A vocabulary of controlled indexing language,
  formally organised, so that relationships between
  concepts are made explicit
• To be used in an information retrieval system,
  such as card catalogue, printed index, or online
  database

3

• The primary purpose of a thesaurus is information
  retrieval
• Secondary purposes
• definition of terms
• semantic map of subject area

4
Classic use of thesaurus

• Thesaurus is used for indexing documents to
  create a database
• The same thesaurus is used for searching the
  database for document retrieval

5
Planning for a thesaurus

• Do you need a thesaurus?
• Controlled language versus natural language
• Indexing in the full-text era

6
Controlled and Natural Language

• Advantages of controlled language
• Eases searching by
• controlling synonyms
• leads to nearest preferred term for starting
  search
• shows relationships between terms
• uses agreed definitions of terms
• Avoids over exhaustivity - no minor terms
• Useful in multilingual systems

7
Weaknesses of controlled language

• lack of specificity
• lack of exhaustivity
• not always up-to-date
• indexer can misunderstand authors intent
• searcher must use un-natural language

8
The thesaurus in the full-text era

• If the system searches for full-text, do you need
  a thesaurus?
• In some systems, the thesaurus enhances the
  performance of full-text retrieval. It retrieves
  extra references that do not contain the exact
  words in the search term.

9
Vocabulary control

• Preferred and non-preferred index terms
• General categories
• Form of index term
• Choice of index terms
• Restriction and clarification of meaning

10
Preferred and non-preferred index terms

• An index term is a representation of a concept
• In a thesaurus, a term must be a Preferred Term
  or a Non-Preferred Term
• The Preferred Term is the chosen term

11
General categories

• Concepts covered by index terms cover three
  categories
• Concrete, e.g. Primates, Buildings
• Abstract, e.g. Evolution, Efficiency
• Individual, e.g. Nigeria, Unesco

12
Form of index term

• Grammatical phrases
• Singular and plural forms
• Punctuation, capitalisation
• Abbreviations, initialisms, acronyms

13
1 Grammatical phrases

• Nouns and noun phrases
• Women workers
• Philosophy of education
• Heads of state
• Adjectives can be used, e.g. Portable
• Verbs, e.g. Communication, Walking
• Initial articles depend upon the sense made, e.g.
  use Theatre, and The Narrows

14
2 Singular and plural forms
15
3 Punctuation, capitalisation

• Avoid punctuation if possible, as it affects
  filing order
• Use brackets only as a qualifier
• Coating (process)
• Retain apostrophes
• Childrens hospitals
• Keep hyphens to retain meaning
• X-Ray

16
4 Abbreviations, initialisms, acronyms

• Use initials if the term or organisation is
  well-known by its initials
• Unesco
• AIDS

17
Choice of index terms

• Use loan words, e.g. Sub judice, glasnost
• Neologisms are ok, e.g money laundering
• Avoid tradenames, e.g. Ovaltine (TM)
• Use popular name or scientific names according to
  the audience, e.g Rubella or German Measles
• Same with place names, e.g. Burma or Myanmar

18
Restriction and clarification of meaning

• Homographs - must clarify meaning
• Cells (biology)
• Cells (electric)
• Reading (process)
• Reading (place)
• Use scope notes and definitions to clarify
  meaning of terms

19
Specificity and Compound Terms

• Vocabulary specificity
• Compound terms levels of pre-coordination
• A thesaurus with mostly single terms is called
  post-coordinate. A thesaurus with many compound
  terms is pre-coordinate

20
1 Vocabulary specificity
21
2 Compound Terms

• It is a general rule that terms in a thesaurus
  should represent single or unitary concepts as
  far as possible, and compound terms should be
  broken into single elements, except when this
  will lower the users understanding
• avoid gt Workload of dentists in Scotland
• prefer gt Workload Dentists Scotland

22
(No Transcript)
23

• Use natural word order,
• e.g. Dried vegetables
• not Vegetables, Dried
• If indirect form is used, there must always be a
  reference to natural word order
• Libraries, University
• see
• University libraries

24
(No Transcript)
25
Keep a compound term to retain the meaning that
would be lost if words were split

• Group discussion
• Discussion group
• Markov Process
• White elephants (which arent elephants at all)
• First aid (would First make a useful search
  term?)

26
Structure and relationships

• A very important feature of a thesaurus is the
  display of structural relationships between the
  terms it contains.
• gt Micro level semantic links between terms
• gt Macro level How the terms relate to the
  overall structure of the subject field

27
Basic relationship types

• Equivalence relationships
• Hierarchical relationships
• Associative relationships

28
1. Equivalence relationships I

• A relationship between preferred and
  non-preferred terms where two or more terms are
  regarded as referring to the same concept
• Use preferred term
• UF (use for) non-preferred term

29
1. Equivalence relationships II

• Popular /scientific names
• Spiders / Arachnida
• Different words for the same concept
• Teleworking / Distance working
• Railways / Railroads
• Words from different languages
• Aliens / Foreigners
• Terms that change over time
• Negroes / Black Americans / African-Americans

30
1. Equivalence relationships III

• Different spelling
• Gipsies / Gypsies
• Mouse / Mice
• Direct and indirect forms
• Dried vegetables / Vegetables, Dried
• Abbreviations and full form
• TQM / Total Quality Management
• Near synonyms
• Dryness / Wetness
• Literacy / Illiteracy

31
2. Hierarchical relationships I

• The relationship between levels of
  superordination and subordination. It shows
  broader and narrower concepts.
• It is an important feature of a thesaurus and
  helps with searching
• BT Broader Term
• NT Narrower Term

32
2. Hierarchical relationships IIa

• Generic
• BT Vertebrates
• NT Mammals
• Reptiles

33
2. Hierarchical relationships IIb

• Whole - part relationships
• BT Ear
• NT External ear
• Inner ear

34
2. Hierarchical relationships IIc

• Instances
• BT Seas
• NT Baltic Sea
• China Sea

35
3. Associative relationships

• Closely related concepts that are not
  hierarchical and not equivalent. The indexer
  often makes a judgement before adding this sort
  of relationship to the thesaurus
• RT Related Term

36
3. Associative relationships - examples

• Earthquakes / seismology
• Accountancy / accountants
• Driving / Road vehicles
• Women / Feminism
• Friction / Wear
• Pests / Pesticides
• Single people / Married people
• Communication / Communication skills

37
Thesaurus Displays

• Can be categorised as
• Alphabetical displays showing scope notes
  hierarchical relationships
• Hierarchical displays generated from A-Z display
• Systematic displays showing total structure
• Graphic displays

38
Construction Techniques 1

• Initials steps
• Define the subject
• Choose thesaurus characteristics and layout
• Select the terms
• other thesauri and classification schemes
• literature searching
• experts knowledge
• Record the terms

39
Construction Techniques 2

• Creating the structure
• organise terms into broad categories
• identify
• key categories
• peripheral categories
• general concepts

40
Construction Techniques 3

• Step A - Choose key facets / groups
• Step B - Make the hierarchies code with NT/BT
• Step C - Add SN, UF, then BT/NT/RT from terms in
  other subject fields
• Step D - Make the alphabetical display