Taxonomies and Indexing: A Technical Strategy

1
Taxonomies and Indexing A Technical Strategy

• Diane Vizine-Goetz
• Office of Research
• OCLC Online Computer Library Center, Inc.

2
Context

• Techniques and approaches developed by for
  libraries and other institutions responsible for
  preserving the human record
• Broad scope
• Long tradition of information organization

3
Why organize information?

• For
• Search and retrieval
• Use
• Preservation disposition

4
Why Organize Information by Subject?

• Find information on a particular subject
• Only and all relevant information
• precision
• recall
• Find related information

5
How?

• Subject analysis
• Conceptual analysis--Determining what an
  information object is about
• Translate concepts into knowledge organization
  (KO) scheme
• e.g., Subject indexes
• Thesauri
• Classification scheme
• Automated, Semi-automated, Human/Intellectual

6
Automation Subject Analysis
7
Automated Concept Identification

• Automated Indexing
• Ranges from simply identifying words in a
  document, to
• Sophisticated analyses that identify key names,
  words, and phrases
• WordSmith Project http//orc.rsch.oclc.org5061/
• Automated Classification
• Automated assignment of documents to categories
  or classes

8
Political News Concepts Extracted by WordSmith

• fair housing
• fair housing act
• family planning
• family planning programmes
• family planning programs
• family planning services
• federal government
• federal government deficit
• federal reserve
• federal reserve bank
• federal reserve board
• federal reserve chairman alan greenspan
  
• federal reserve system

9
(No Transcript)
10
(No Transcript)
11
(No Transcript)
12
(No Transcript)
13
Advantages of automatic concept identification

• Inexpensive
• Suitable for indexing/categorizing large
  quantities of text
• Can identify popular and emerging concepts and
  terminology

14
Why use knowledge organization schemes?

• Knowledge organization schemes such as subject
  heading lists, thesauri, classification schemes
  are specialized languages designed for retrieving
  information
• Goal--to reduce ambiguities that cause precision
  recall failures

15
Free text v.s. controlled subject retrieval
language

• WordSmith
• family planning
• family planning programmes
• family planning programs
• family planning services

• Library of Congress Subject Headings (LCSH)
• Birth control clinics
• UF Family planning services
• Planned parenthood services
• BT Clinics
• 19860211

16
MeSH Heading vs. LCSH

• Family Planning
• Note Programs or services designed to assist
  the family in controlling reproduction by either
  improving or diminishing fertility.
• Entry Term
• Birth Control
• Planned Parenthood
• Basal Body Temperature Method
• Birth Limiting
• Births Averted
• Family Planning Surveys
• ...

• Birth control (19880919)
• UF Family planning
• Planned parenthood
• Population control
• Pregnancy--Prevention
• BT Hygiene, Sexual
• Sexual ethics
• RT Contraception
• Family size
• NT Abortion
• Birth Intervals
• Childlessness
• ...

17
Characteristics of subject retrieval languages

• Terminology is often domain specific
• Medicine gt MeSH Engineering gt INSPEC
  Agriculture gt Agrovoc
• Control vocabulary (synonyms homonyms)
• Express relationships between terms

18
Within a domain, terms are context independent

• Ei Thesaurus
• TM
• Bank protection
• UF
• Coastal engineering--Bank protection
• Inland waterways--Bank protection
• SN
• Protection of river banks and lake shores. For
  seacoasts, use SHORE PROTECTION
• DT January 1993

• BT
• Protection
• RT
• Banks (bodies of water)
• Coastal engineering
• Environmental engineering
• Erosion
• Inland waterways
• River control
• Shore protection
• Slope protection
• Soil conservation
• MC 407.2 407.3
• OC 914.1

19
Controlled Vocabulary

• Preferred way of expressing a concept
• e.g., Popular vs. technical
• Heart attack vs. Myocardial infarction
• Non-used vocabulary often included
• Synonyms
• Current/Outdated terms gt Disabled/Handicapped
• Lexical variants
• Phrase/Inverted forms gt Bilingual
  education/Education, Bilingual
• Quasi-Synonyms
• Synonyms/Antonyms gt Literacy/Illiteracy

20
Relationships

• Equivalence
• Synonymous terms
• Hierarchy
• Generic relationship (kind)
• Whole-part relationship
• Instance relationship (example)
• Association

21
Subject Retrieval using a controlled vocabulary
22
Related Terms in LCSH
23
Classification / Categorization System

• A systematic arrangement of knowledge into useful
  categories
• General schemes special schemes
• DDC, LCC, UDC AGRIS, MSC
• Present a generalized view of knowledge at
  varying levels of depth
• May be enumerative or synthetic

24
Some Advantages of Traditional Schemes

• Meaningful notation
• Well-developed hierarchies
• Well-defined categories
• Rich network of relationships

25
Meaningful Notation (DDC)

• 005.1 Programming
• 005.1 Programmation
• 005.1 ????????????????
• 005.1 Programación

26
DDC Notation Indicates Hierarchy

• 600 Technology
• 630 Agriculture
• 633 Field and plantation crops
• 633.1 Cereals
• 633.11 Wheat
• 633.12 Buckwheat
• 633.13 Oats

27
Well-developed Hierarchies
28
Hierarchies Categories

• Hierarchical from general to specific
• Categories have superordinate, coordinate,
  subordinate relationships in hierarchy
• Subcategories must be mutually exclusive

29
Hierarchies Categories

• Top gt Recreation gt Automotive gt Driving gt Road
  Rage
• Social Problems gt Public Safety gt Traffic Hazards
  gt Highways gt Road Rage

30
Hierarchies, Categories, Relationships

• 500 Science
• 510 Mathematics
• 512 Algebra, number theory
• 512.3 Fields
• Class here field theory, Galois theory
• Class linear algebra in 512.5 class
  number theory in 512.7

31
Advantages of Category Schemes

• Facilitate retrieval based on concepts not simply
  keywords
• Provide context for search terms (disambiguates)
• Facilitate browsing search refinement

32
(No Transcript)
33
(No Transcript)
34
(No Transcript)
35
(No Transcript)
36
Advantages Disadvantages of Formal KO Schemes

• Bring like items together
• Provide context show relationships
• Support browsing
• May accommodate multilingual usage
• -
• Reactive to emerging topics
• Terminology may not match users
• Not practical to apply to everything

37
Advantages Disadvantages of Free Text

• Latest terminology
• Application not an issue
• -
• User must to produce synonyms and relationships
• Limited browsing
• Little multilingual support

38
Other Solutions

• Combine approaches
• Map among KO schemes
• Map free text terms to KO schemes
• Produce supplemental browsable indexes from free
  text

39
Resources

• ANSI/NISO Z39.19-1993 (Revision of ANSI
  Z39.19-1980) Guidelines for the Construction,
  Format, and Management of Monolingual Thesauri
  lthttp//www.niso.org/stantech.htmlz3919gt
• Controlled vocabularies, thesauri and
  classification systems available in the WWW. DC
  Subject lthttp//www.lub.lu.se/metadata/subject-hel
  p.htmlgt
• The Intellectual Foundation of Information
  Organizationby Elaine Svenonius. MIT Press
  ISBN 0262194333
• List of Web Subject Resources lthttp//www.loc.gov/
  catdir/pcc/saco/resources.htmlgt
• The Organization of Information (Library and
  Information Science Text Series) by Arlene G.
  Taylor. Libraries Unlimited ISBN 1563084988
• Resources for Indexers lthttp//www.asindexing.org/
  asires.shtmlgt