Manifestation of Research Specialty Processes in Collections of Journal Papers

1
Manifestation of Research Specialty Processes in
Collections of Journal Papers

• Steven A. Morris
• Oklahoma State University

2
Summary

• Discuss research specialties
• Model collections of papers as systems of coupled
  bipartite networks
• Discuss entities, links, and entity groups as
  manifestations of research specialties in
  collections of papers
• Discuss visual presentation to reveal structural
  and dynamic information about a specialty

3
Goals

• Visualize structure and dynamics of a research
  specialty through a collection of papers
• Social organization
• Knowledge organization
• Present to subject matter experts for technology
  forecasting

4
Research specialty definitions

• A research specialty is a self-organized social
  organization whose members tend to study a common
  research topic, attend the same conferences,
  publish in the same journals, cite each other's
  work, and belong to the same social networks that
  are known as invisible colleges.
• Specialties create their own literature, i.e., a
  body of journal papers and books that broadly
  focus on the specialty's research topic.
• Define a collection of papers as a list of
  journal papers that constitutes a comprehensive
  sample of a specialty's journal literature.

5
Model of a research specialty
Generated knowledge adopted as base knowledge
produces paradigm creep
Funding
Technical communication through journal
literature and conferences
Base knowledge
Research reports
Body of knowledge
Kuhnian paradigm
Researchers

• Researcher local organization(Researcher team
  processes)
• Researcher global self-organization(Research
  global communication processes)
• Researcher education training(Researcher
  entrance processes)
• Researcher retirement/out-migration(Reseacher
  exit processes)

• Symbolic generalizations
• Metaphysical paradigms
• Validation standards
• Exemplars

• Journal literature
• Conference literature
• Educational theses dissertations
• Institutional reports
• Books

6
Size of specialties

• Specialties are usually small, less than 100 core
  members according to Kuhn.
• Collections of papers usually less than 5000
  papers.
• Scaling not a big problem.

7
Static information sought about a specialty

• Identification and ranking of individual entities
• Experts
• Productive researchers
• Rising stars
• Centers of excellence
• Exemplar references
• Key journals

8
Static information sought about a specialty

• Structural mapping (groups and their relations)
• Terms (subtopic vocabularies)
• Papers (research fronts papers grouped by
  subtopic)
• References (exemplar reference groups,
  paradigms)
• Paper authors (research teams)
• Reference authors (schools of thought)
• Paper journals (research report libraries)
• Reference journals (base knowledge libraries)

9
Dynamic information sought about a specialty

• Monitoring
• Trends
• Growth/decline of the specialty
• Obsolescence of knowledge
• Geographic migration of research activity
• Discontinuous events
• Discoveries
• External events
• Forecasting
• Extrapolate trends
• Predict risk of discontinuous events

10
Why use journal papers to investigate a specialty?

• Vetted through review process
• Public record of researcher communication
• Permanent record
• Formatted, structured information available
  (through abstract services)

11
Gathering collections of papers to cover a
specialty

• Gathered from Science Citation Index
• Using seed references
• Find all papers that cite a collection of key
  references in the specialty
• Query of terms.
• Find all papers associated with keyword terms
  that are related to the specialty
• Index terms, title terms, abstract terms
• Query of reference authors
• Find all papers that reference key authors in the
  specialty.

12
Entity-Relation model of a collection of journal
papers
authors
reference authors

• 6 direct bipartite networks
• 15 indirect bipartite networks formed from
  cascading bipartite networks

index terms
references
papers
paper journals
reference journals
13
Entity-relationship model of a collection of
journal papers
Other entities Paper year Reference year
REFERENCE AUTHORS
PAPER AUTHORS
APPEARS ONCE IN MULTIPLE
HAS MANY UNIQUE
APPEARS ONCE IN MULTIPLE
HAS ONE
APPEARS MULTIPLE TIMES IN MULTPLE
CONTAINS MULTIPLE UNIQUE
REFERENCES
INDEX TERMS
PAPERS
CONTAINS MULTIPLE MULTPLE TIMES
APPEARS ONCE IN MULTIPLE
APPEARS ONCE IN MULTIPLE
APPEARS IN ONE
HAVE MANY UNIQUE
HAS ONE
REFERENCE JOURNALS
PAPER JOURNALS
citing entities
cited entities
14
Bibliometric entities vs. physical entities
Paper author H. G. Small
Reference author Small HG
Reference author Small H
Bibliometric entities are objects in the paper
collection and acquire separate meaning.
Physical entities are objects in the real world
that correspond to bibliometric entities.
Physical author Henry Small
15
Entity-relationship diagram showing relation of
physical entities to bibliometric entities
F404_2
16
Papers citing papers networks
paper
citation
17
Bibliographic entities as tokens of research
specialty objects
REFERENCE AUTHORS
Base knowledge generators, experts
PAPER AUTHORS
Researchers
Research reports
REFERENCES
TERMS
Concept symbols, base knowledge
PAPERS
Concept symbols
Base knowledge archives
REFERENCE JOURNALS
PAPER JOURNALS
Report archives
18
Bibliographic links as tokens of research
specialty relations
REFERENCE AUTHORS
PAPER AUTHORS
Researcher generated base knowledge represented by
Researcher participated in research reported by
Term associated with research reported by
REFERENCES
TERMS
PAPERS
Research reported used base knowledge represented
by
Journal archives research reported by
Journal archives base knowledge represented by
REFERENCE JOURNALS
PAPER JOURNALS
19
Networks in collections of journal papers
Like entities entities of same
entity-type Unlike entities entities drawn from
more than one entity-type
authors
papers
Bipartite network
Unipartite cooccurence network
20
Networks in collections of journal papers
authors
reference authors
terms
references
papers
paper journals
reference journals
21
Cascaded bipartite networks
22
Occurrence and co-occurrence matrices
Each occurrence matrix has two associated
co-occurrence matrices.
23
Co-occurrence relations
Co-occurrence relations are used map the
structure of a scientific specialty by providing
a means to find entity groups through clustering.
24
Bibliographic cooccurrence links as tokens of
unipartite research specialty relations
Two researchers both generated base knowledge
used by research reported by
REFERENCE AUTHORS
PAPER AUTHORS
Two researchers worked together on research
reported by
Two papers associated with similar research
Two pieces of base knowledge both used in
research reported by
REFERENCES
TERMS
PAPERS
Two papers research both used base knowledge
represented by
Two terms both associated with research reported
by
REFERENCE JOURNALS
PAPER JOURNALS
25
Entity feature vectors
Features are measurable quantities used to
characterize entities for pattern recognition and
clustering purposes. A feature vector is an array
of features used for pattern recognition and
clustering. Each entity has two feature vectors
per occurrence matrix.
Paper author to reference author matrix, Oapar
Paper author co-occurrence matrix, Capar
Paper author i
Occurrence feature vector, Oiapar
Co-occurrence feature vector, Ciapar
F404_34
26
Interpretation of occurrence feature vectors
Examples of occurrence feature vectors for
entities in a collection of papers.
Primary entity-type x1 Secondary entity-type x2 Feature vector for entity i Interpretation
paper reference Oipr a) The concept symbols used by a paper (Small, 1978). b) the knowledge sources used by a paper.
reference paper Oirp The papers using a reference as a concept symbol.
paper author paper Oiapp A paper authors oeuvre
paper author reference author Oiapar The reference authors whose work a paper author reads and uses. An authors identity (White, 2001).
reference author paper author Oiarap The paper authors that read and use a reference authors work.
paper journal reference journal Oijpjr The reference journals holding source knowledge used by papers in a paper journal
reference journal paper journal Oijrjp The paper journals whose papers draw knowledge from a reference journal
paper terms Oipt A papers research vocabulary
27
Interpretation of co-occurrence feature vectors
Examples of co-occurrence feature vectors for
entities in a collection of papers.
Primary entity-type x1 Secondary entity-type x2 Feature vector for entity i Interpretation
paper reference Cipr The papers that use the same concept symbols as paper i. (Papers covering the same topic as paper i.)
reference paper Cirp The references being used by the same papers the use reference i. (Exemplar references for the same Kuhnian paradigm as reference i. )
paper author paper Ciapp The collaborators of paper author i.
paper author reference author Ciapar The paper authors using the same knowledge sources as paper author i. Paper author is invisible college.
reference author paper author Ciarap The reference authors used as knowledge sources by the same paper authors as reference author i. The image of reference author i. (White, 2001)
paper journal reference journal Cijpjr The paper journals using the same sources of knowledge as paper journal i.
reference journal paper journal Cijrjp The reference journals (sources of knowledge) being used by the same paper journals as reference journal i.
paper terms Cipt Papers using the same research vocabulary as paper i. (Papers covering the same topic as reference journal i.)
28
Bibliographic cooccurrence clusters as tokens of
research specialty group objects
REFERENCE AUTHORS
PAPER AUTHORS
Base knowledge generator groups. schools of
thought
Research teams
Base knowledge groups, paradigms
Research team oeuvres
REFERENCES
TERMS
PAPERS
Research front papers by topic
Research subtopic vocabularies
Research front library
Base knowledge libraries
REFERENCE JOURNALS
PAPER JOURNALS
29
Visualization of matrices
30
Research front timeline
31
Major sub-specialies
32
Historical development
33
Research front to reference crossmap
CURRENT TOXIN RESEARCH
EARLY TOXIN RESEARCH
CURRENT VACCINE RESEARCH
ANTHRAX DETECTION
EARLY RESEARCH TOXIN AND VACCINE
BIOTERROR
1980S EARLY 90S TOXIN AND VACCINE RESEARCH
34
Reference usage plot
BIOTERROR
OLD REFERENCES STILL CURRENT
DETECTION
BREAKTHROUGH IN TOXIN RESEARCH
LEPPLA FRIEDLANDER KEY REFERENCES IN TOXIN
RESEARCH
BRACHMAN STUDY OF VACCINE EFFICACY
EARLY INHALATION ANTHRAX RESEARCH
OBSOLETE EARLY RESEARCH
35
Paper author usage plot
MOCK
COLLIER
TURNBULL
FRIEDLANDER
LEPPLA
NO LONGER ACTIVE
NO LONGER ACTIVE
THORNE
WRIGHT
36
Research front to index terms crossmap
TOXIN EXPRESSION TERMS
TOXIN TERMS
VACCINE TERMS
BIOTERROR TERMS
37
Questions
38
PROPOSED ENTITY-RELATIONSHIP MODEL FOR TERRORIST
INCIDENT REPORTS
TERROR-IST
TERROR-IST GROUP
LINGUIS-TIC TERMS
LAW INFORCE-MENT OFFICER

• OTHER ENTITIES
• EVENT DATE
• REPORT DATE

EVENT REPORT
INCIDENT TYPE
GOVERN-MENT OFFICIAL
PERSONAL NAMES
VICTIM
ENTITIES IN YELLOW TO BE IMPLEMENTED
TOWN
DISTRICT
COUNTRY
DIRECT LINKS FROM ENTITY EXTRACTION
LINKS SUPPLIED BY ANALYSTS OR INFERENCE
39
(No Transcript)