Domain analysis An important part of thesaurus construction Marianne Lykke Nielsen The Royal School

1
Domain analysisAn important part of thesaurus
construction Marianne Lykke NielsenThe Royal
School of Library and Information
ScienceInstitute of Information
Studiesmln_at_db.dk ASIS SIG CR
WorkshopNovember 12, 2000
2

• Outline
• Motivation
• Perspective of domain analysis
• Case study
• Findings and implications for thesaurus design
• Study methods
• Summary

3
Motivation Knowledge of the clientele to be
served, their problems and needs, and of the
search requests to be expected is extremely
important in the design of every function of the
information system (Soergel, 1985) Indexing and
information system design needs to proceed within
the context of a deep understanding of the
character of the intellectual domain, its
culture, and its research questions (Bates,
1998)
4
Perspective of domain analysis

(Ingwersen, 1996)
INFORMATION OBJECTS- Text/Knowledge
representations - Full text, pictures/Semantic
entities ? Models ? Interface/ Individual
users Intermediary COGNITIVE SPACE - Work
task/Interest Social/org. environment -
Current Cognitive State - Domains
Query Request Models Models functions ?
Models ? - Problem/Goal - Strategies/Goals -
Uncertainty - Tasks Preferences -
Information need - Information behaviour
IR SYSTEM SETTING - Search language/IR
techniques cognitive transformation -
Database structure and influence-
Indexing rules/computational logic
interactive communication? Models ?
of cognitive structures
5

• Case study
• The domain analysis has focused on investigating
  the following factors
• the nature of the work domains and the
  professionals
• work tasks and derived information needs
• information use
• professional background
• language use
• searching behavior and search problems
• the subject field/discipline
• the literature
• the available resources for indexing and
  thesaurus construction

6

• Findings and implications for thesaurus design
• The domain analysis provided the needed knowledge
  about the environment of the thesaurus and made
  it possible
• to define the role of the thesaurus
• to define the topical focus of the thesaurus
• to define viewpoint and approach
• to define the structure of the thesaurus
• to choose sources for thesaurus construction

7
Lu 21-098 GN Irindalone BT
Antihypertensive agents Lundbeck
compounds RT Concomitant medication
Depression Hypertension
Serotonin uptake inhibitor SN Activity
Enhances the effect of Serotonin uptake
inhibitors, Serotonin (5-HT2C) antagonist
SC C1 Lu drugs
8
Animals UF Experimental animals
Laboratory animals BT Biological test
systems NT Fishes Gene
manipulated animals Insects
Mammals RT Species SC A Organisms and
test systems
9
Open clinical trials UF Open clinical
studies Open label studies
Open pilot studies Open studies
Open trials BT Types of clinical trial
control SN A trial in which subjects and
investigators know which product each
subject is receiving opposite of double-
blind study. (Applied Clinical Trials) SC
P Management and administration
10

• Study methods
• Group interviews
• Content analysis and discourse analysis of user
  requests
• Word association test

11

• Summary
• The development of a thesaurus should be based
  on a thorough understanding of the environment of
  the thesaurus
• Two qualitative and two quantitative methods
  were tested. The qualitative methods produce
  domain knowledge, while the quantitative methods
  mostly provide conceptual knowledge
• The qualitative and quantitative methods
  complement each other
• The findings mostly represent the work domain
• The user must interact actively with the
  knowledge structure to bridge the gab between the
  work domain and his own individual perceptions,
  tasks, and information needs.

12

• Group interviews
• 8 group interviews involving 29 participants
• Structured, but open interviews
• Follow-up questionnaire to avoid dominance and
  get some quantitative data
• Interviews were taped and summarized in short
  interview description
• Provided a general knowledge about
• work tasks, information use, seeking and
  searching behavior
• Back

13

• Content analysis of user requests
• 50 written user requests, sent by e-mail to the
  Library were analyzed
• Types and numbers of categories
• Relations between concepts types
• Terminology
• Provided an overview over topical focus,
• important facets and their relationships
• Back

14

• Discourse analysis of user requests
• 50 written user requests, sent by e-mail to the
  Library were analyzed
• Importance of concepts
• Approach and perspective
• Understanding and use of concepts
• Provided a detailed understanding of work tasks
  and related
• information use, naming and use of concepts and
  terms
• Back

15

• Word association test
• Pilot test of 24 stimulus words, 3 respondents
• Instruction to make associations according to
  work tasks and practical perspective
• Understanding and use of concepts
• Provided work-task related relationships,
  work-task oriented synonyms, and diversity of
  naming
• Example
• Back

16

• Roles
• Searching thesaurus supporting work-task related
  searching in local and external information
  systems. Weight on synonyms (to expand search
  queries) and RT (to analyze search request and
  formulate search query)
• Indexing thesaurus supporting controlled, manual
  indexing of local information systems
• Glossary of important, commonly used concepts
  and terms. Weight on definitions, understandings
  and use of terms
• Back

17
Work task groups (discourses)
Back
18
Example from word association test Electrocardiog
rams UF EKG (LPH) RT Q-T Prolongation (MBS)
Sertindole (MBS) Side effects
(MBS) Heart function (EDC) Heart
rythm (LPH) Safety pharmacology (EDC)
Back