Knowledge Synthesis

3 Structures

Structures of Knowledge

Drawing a big picture of scientific knowledge has always been

desirable for various reasons

Science mapping attempts to find representations of intellectual

connections within the dynamically changing system of scientific

knowledge (Small, 1997)

Science mapping aims at displaying the structural and dynamic

aspects of scientific research (Börner et al.2003; Morris et al., 2008)

Science mapping uses mainly the “structures of knowledge”

Three Structures of Knowledge

Discovering hidden patterns

SOCIAL

STRUCTURE

how authors, institutions and countries

interact each other

CONCEPTUAL

STRUCTURE

what science talks about, the main

themes and trends

INTELLECTUAL

STRUCTURE

CONCEPTUAL

STRUCTURE

Each community of scientists would have a complete overview of the main findings related

to their specific field, following the evolution of theories and techniques

Science Mapping allows investigating scientific knowledge from a statistical point of view

how the work of an author influences

a given scientific community

Knowledge Synthesis

Conceptual Structure

Conceptual Structure Tab

Conceptual structure represents relations among concepts or words in a set of

publications:

•Words, which appear together in a document, will be related in a network. It

is also known as the co-words network. This structure is be used to

understand the topics covered by a research field to define what are the most

important and the most recent issues (so called, research front). It could also

help in the study of the evolution of subjects over time.

•Similarly, to network analysis, factorial analysis (data reduction techniques) is

helpful in identifying subfields. Various dimensionality reduction techniques

can be applied, such as correspondence analysis (CA), multiple

correspondence analysis (MCA), multidimensional scaling (MDS), principal

component analysis (PCA). Clustering algorithms can be used in both cases of

network or factorial analysis.

•Mixed approach. Starting from a conceptual network, you identify thematic

networks that plot on abi-dimensional matrix, where axis are function of

centrality and density of the thematic network. Dividing the timespan in time

slices, it is possible to represent the thematic evolution within a specific

research field through an alluvial graph.

Mapping conceptual structure

The scope of a science mapping study can be a scientific discipline, a field of

research, or topic areas concerning specific research questions

keywords

documents

Factorial approach:

Correspondence Analysis

Multiple Correspondence Analysis

Multidimensional Scaling

Network approach

Co-word analysis

keywords

Conceptual Structure Tab

Factorial Approaches

Conceptual Structure Tab

The basic idea behind factorial approaches is to reduce the

dimensionality of data and represent it in a low-dimensionality space.

Three alternative methodologies:

Correspondence Analysis (CA)

Multiple Correspondence Analysis (MCA)

Multidimensional Scaling (MDS)

Factorial Approach: Interpretation

Conceptual Structure Tab

The proximity between words corresponds to shared-substance:

keywords are close to each other because a large proportion of articles treat

them together;

they are distant from each other when only a small fraction of articles uses

these words together.

The origin of the map represents the average position of all column

profiles and therefore represents the center of the research field

(meaning common and large shared topics) (Cuccurullo et al.2016)

Correspondence Analysis and Clustering

Map of words

Conceptual Structure Tab

Factorial Analysis

Each color

represents a

cluster of

word (a

“topic”)

Clusters are

identified by

hierarchical

clustering

Correspondence Analysis and Clustering

Dendrogram of words

Conceptual Structure Tab

Factorial Analysis

The height

measures the

distance

among words

or cluster of

words

“Similar

words” which

explain a

similar

“concept” or

“topic”

“distant words”

which define

different

“concept”

or “topic”

The height helps to

choose where to cut

the dendrogram

defining the partition

Correspondence Analysis and Clustering

Most contributing documents

Conceptual Structure Tab

Factorial Analysis

The map plots the

documents

associated to the

highest absolute

contribution

The absolute

contributions

measure the

weight of each

document in the

information

summarized by

the two axes

The colors

represent the

clusters to which

each document

belongs

This graph allows

to identify the

link among topic

and documents

Correspondence Analysis and Clustering

Most cited documents

Conceptual Structure Tab

Factorial Analysis

Network Analysis

Conceptual Structure Tab

Co-occurrence Network

Graph theory is the study of graphs, which are mathematical

structures used to model pairwise relations between objects

A graph is made up of vertices (also called nodes or points)

which are connected by edges (also called links or lines)

A distinction is made between undirected graphs, where edges

link two vertices symmetrically, and directed graphs, where

edges, then called arrows, link two vertices asymmetrically

Network Analysis and Science Mapping

In Science mapping, a network graph is used to represent co-occurrences

among bibliographic meta-data

The starting point is a co-occurrence matrix

𝐼1𝐼2𝐼3𝐼𝑃

𝐼𝑝

… …

𝐼1

𝐼2

𝐼3

𝐼𝑃

𝐼𝑝

…

𝑛11

𝑛22

𝑛𝑝𝑝

𝑛𝑃𝑃

𝑛32

Diagonal elements are the

occurrence of each item in the

collection

e.g in co-word analysis:

•Each items is words and

•an occurrence is the number of

appearances of a particular word in

the document collection.

Non-diagonal elements are

the co-occurrence of two

items in the collection

e.g in co-word analysis:

𝑛32 measures how many times

the words 𝐼3and 𝐼2appears

together in the same corpus

(keyword list, title, abstract, etc.)

Conceptual Structure Tab

Co-occurrence Network

Conceptual Structure Tab

Co-occurrence Network

science

Know.

manag.

perf.

impact

bibliom.

innov. tech.

cit.anal.

int.str.

science

100 17 16 21 20 25 40 26 10 0

knowledge

17 32 0 0 0 0 0 0 0 0

management

16 034 0 0 0 0 0 0 0

performance

21 0 0 70 15 0 0 0 0 0

impact

20 0 0 15 60 014 0 0 0

bibliometrics

25 0 0 0 0 38 13 0 0 0

innovation

40 0 0 0 14 13 80 0 0 0

technology

26 0 0 0 0 0 0 28 0 0

citation

analysis

10 0 0 0 0 0 0 0 40 14

intellectual

structure

0 0 0 0 0 0 0 0 14 36

The vertex size is proportional to the item occurrence (diagonal

elements)

The edge size is proportional to item co-occurrences (non-diagonal

elements)

Each vertex represents an item (in this example, each edge is a word)

Network Analysis and Science Mapping

from the co-occurrence matrix to the network graph

Vertex centrality vs

Vertex farness

Strength

of edges

Vertex

Clustering

(community

detection)

Conceptual Structure Tab

Co-occurrence Network

Network Analysis and Science Mapping

How to read a network graph

Co-occurrence Network

Conceptual Structure Tab

Co-occurrence Network

Each cluster can

be seen as a

“topic”

The colors represent

the clusters to which

each word belongs

Co-occurrence Network

Conceptual Structure Tab

Co-occurrence Network

Network options

Layout

It is possible to choose among

several network layout (Holten et

al., 2009).

“Automatic Layout” automatically

chooses the best layout in terms of

graph readability

Normalization

Co-occurrences can be normalized

by using similarity measures such

as Salton's Cosine, Jaccard's Index,

Equivalence Index, and Association

Strength (van Eck et al.,2009).

Clustering

Several clustering algorithm

are proposed. The best is

“Walktrap”(Lancichinetti et

al., 2009).

Repulsion Force

Repulsion force varies

between 0 and 1, where 1

represents the maximum

separation between the

groups.

Thematic Map

Conceptual Structure Tab

Thematic Map

By applying a clustering algorithm on the keyword network, it is possible to highlight the

different themes of a given domain

Each cluster/theme can be represented on a particular plot

known as Strategic or Thematic map (Cobo et al., 2011):

Centrality is a measure of the theme’s relevance

Density is a measure of the theme’s development

Limitations:

Each keyword is associated only with one theme

It is not possible to use themes for document categorization

It is not possible to jointly analyse meta-information

Callon

centrality

Callon

density

MOTOR THEMES

BASIC AND TRANSVERSAL

THEMES

EMERGING OR DECLINING

THEMES

HIGHLY DEVELOPED AND

ISOLATED THEMES

Thematic Map

from a network to a bivariate map

Conceptual Structure Tab

Thematic Map

Callon

centrality

Callon

density

MOTOR THEMES

BASIC AND TRANSVERSAL

THEMES

EMERGING OR DECLINING

THEMES

HIGHLY DEVELOPED AND

ISOLATED THEMES

Word Co-occurrences

Network

Conceptual Structure Tab

Thematic Map

Each bubble represents

a network cluster

The bubble names are

words, belonging in the

cluster, with the higher

occurrence value

The bubble size is

proportional to the cluster

word occurrences

The bubble position is

set according to the

cluster Callon centrality

and density

Topics in business and management

literature using bibliometrics

Thematic Map

Cluster composition

Conceptual Structure Tab

Thematic Map

Probability of belonging to a cluster

Conceptual Structure Tab

Thematic Map

Papers with a probability of 0.8 or higher almost certainly belong to the cluster k.

Papers with a probability between 0.4 and 0.8 have a high probability of belonging to the cluster k.

A paper can deal with several topics, so it can also belong to more than one cluster.

Thematic Evolution

A longitudinal thematic map analysis

Conceptual Structure Tab

Thematic Evolution

Dividing the time span in different time slices, it is possible to study and

plot the topic evolution (in terms of trajectory along time)

In this graph, the bubble represents an emerging topic that moves towards

mainstream themes area

Callon

centrality

Callon

density

Callon

centrality

Callon

density

Time slice 1 Time slice 2 Time slice t

………

Thematic Evolution

Setting time slices

Conceptual Structure Tab

Thematic Evolution

•1985 –2006

•2007 –2014

•2015 –2021

Van Raan AFJ, 2006 Burst in publications

Looking at the distribution of publication per year, we decided to spit

our collection into 3 time slices setting 2 cutting points 2006 and 2014:

Thematic Evolution

A longitudinal thematic map analysis

Conceptual Structure Tab

Thematic Evolution

Time slice

1985 -2006

Thematic Evolution

A longitudinal thematic map analysis

Conceptual Structure Tab

Thematic Evolution

Time slice

2007 -2014

Thematic Evolution

A longitudinal thematic map analysis

Conceptual Structure Tab

Thematic Evolution

Time slice

2015 -2021

Thematic Evolution

A longitudinal thematic map analysis

Conceptual Structure Tab

Thematic Evolution

it is also possible to highlight the tendencies of some topics to merge

together, or of a topic to split into several themes

Callon

centrality

Callon

density

Callon

centrality

Callon

density

Callon

centrality

Callon

density

Time slice 1 Time slice 2 Time slice t

………

Thematic Evolution

A longitudinal thematic map analysis

Conceptual Structure Tab

Thematic Evolution

Knowledge Synthesis

Intellectual Structure

Intellectual Structure Tab

Intellectual structure shows relationships between nodes which

represent references

Network edges can have different interpretations depending on the

citation type (co-citation or direct citation)

Citation analysis is the most common analysis in bibliometrics in the

form of co-citation between authors or documents.

Co-citation analysis, when examined over time, is helpful in detecting

a shift in paradigms and schools of thought.

Intellectual Structure

Intellectual Structure Tab

Two type of analysis:

Co-citation network

•Small, 1973

Historiographic mapping

•Garfield, 2004

Co-citation analysis

Intellectual Structure Tab

We talk about co-citation of two documents when

both are cited in a third document

Document “A” and “B” are “references” co-cited by

documents “C”, “D” and “E”

Co-citations can be represented in aco-occurrence

matrix just like co-word analysis

Local citations

of 𝐼𝑝

Co-citations

among 𝐼2and 𝐼3

Co-citation network

Intellectual Structure Tab

Co-citation network

Intellectual Structure Tab

Co-citation network

Intellectual Structure Tab

Co-citation network

Historiographic mapping

Intellectual Structure Tab

time

presentpast

Each historical path identifies a research topic and its core

authors/documents

First path

Second path

Historiographic mapping

Intellectual Structure Tab

Some basic concepts about the historiograph:

Each node represents a document (included in the analyzed

collection) cited by other documents

Each edge represents a direct citation (e.g. D cited A; G cited D; etc.)

Nodes and Edges are plotted on an oriented graph where the

horizontal axis represents the publication years

Historiographic mapping

Intellectual Structure Tab

Historiograph

Only one path

Historiographic mapping

Intellectual Structure Tab

Historiograph

Only one path

Historiographic mapping

Intellectual Structure Tab

Historiograph

Authors’ keywords Keywords Plus

Historiographic mapping

Intellectual Structure Tab

Historiograph

Local

citations

Global

citations

Knowledge Synthesis

Social Structure

Social Structures

Social structure shows how authors or institutions relate to others in

the field of scientific research.

The most common kind of social structure is co-authorship network

(Peters et al.1991)

With co-authorship networks can be discovered, for example, groups

of regular authors, influent authors, hidden communities of authors,

relevant institutions in a specific research field, etc.

Social Structure Tab

Author collaboration

Social Structure Tab

Collaboration Network

Institutions collaboration

Social Structure Tab

Collaboration Network

Country collaboration

Social Structure Tab

Collaboration Network

Country collaboration

Social Structure Tab

Collaboration World Map

References (1)
•Aria, M., & Cuccurullo, C. (2017). bibliometrix: An R-tool for comprehensive science mapping analysis. Journal of Informetrics, 11(4), 959-975.
•Aria M., Misuraca M., Spano M. (2020) Mapping the evolution of social research and data science on 30 years of Social Indicators Research, Social 
Indicators Research. (DOI: https://doi.org/10.1007/s11205-020-02281-3)
•Aria, M., Cuccurullo, C., D’Aniello, L., Misuraca, M., & Spano, M. (2022). Thematic Analysis as a New Culturomic Tool: The Social Media Coverage on 
COVID-19 Pandemic in Italy. Sustainability, 14(6), 3643, (https://doi.org/10.3390/su14063643)
•Aria M., Alterisio A., Scandurra A, Pinelli C., D’Aniello B, (2021) The scholar’s best friend: research trends in dog cognitive and behavioural
studies, Animal Cognition. (https://doi.org/10.1007/s10071-020-01448-2)
•Cuccurullo, C., Aria, M., & Sarto, F. (2016). Foundations and trends in performance management. A twenty-five years bibliometric analysis in business 
and public administration domains, Scientometrics, DOI: 10.1007/s11192-016-1948-8 (https://doi.org/10.1007/s11192-016-1948-8)
•Cuccurullo, C., Aria, M., & Sarto, F. (2015). Twenty years of research on performance management in business and public administration domains. 
Presentation at the Correspondence Analysis and Related Methods conference (CARME 2015) in September 2015 
(https://www.bibliometrix.org/documents/2015Carme_cuccurulloetal.pdf)
•Sarto, F., Cuccurullo, C., & Aria, M. (2014). Exploring healthcare governance literature: systematic review and paths for future 
research. Mecosan (https://www.francoangeli.it/Riviste/Scheda_Rivista.aspx?IDarticolo=52780&lingua=en)
•Cuccurullo, C., Aria, M., & Sarto, F. (2013). Twenty years of research on performance management in business and public administration domains. 
In Academy of Management Proceedings (Vol. 2013, No. 1, p. 14270). Academy of Management (https://doi.org/10.5465/AMBPP.2013.14270abstract)
•Belfiore, A., Salatino, A., & Osborne, F. (2022). Characterising Research Areas in the field of AI. arXiv preprint 
arXiv:2205.13471.(https://doi.org/10.48550/arXiv.2205.13471)
•Belfiore, A., Cuccurullo, C., & Aria, M. (2022). IoT in healthcare: A scientometric analysis. Technological Forecasting and Social Change, 184, 122001. 
(https://doi.org/10.1016/j.techfore.2022.122001)

References (2)

•Börner, K., Chen, C., & Boyack, K. (2003). Visualizing knowledge domains. Annual Review of Information Science and Technology, 37,179–255.

•Cobo, M. J., López-Herrera, A. G., Herrera-Viedma, E., & Herrera, F. (2011). An approach for detecting, quantifying, and visualizing the evolution of a

research field: A practical application to the fuzzy sets theory field. Journal of Informetrics, 5(1), 146-166

•Garfield, E., & Sher, I.H. ( 1993). Keywords PlusTM algorithmic derivative indexing. Journal of the American Society for Information Science,44( 5), 298–

299.

•Garfield, E. (2004). Historiographic mapping of knowledge domains literature. Journal of Information Science,30(2), 119-145.

•Holten, D., & Van Wijk, J. J. (2009, June). Force‐directed edge bundling for graph visualization. In Computer graphics forum (Vol. 28,No. 3, pp.983-

990). Oxford, UK: Blackwell Publishing Ltd.

•Lancichinetti, A., & Fortunato, S. (2009). Community detection algorithms: a comparative analysis. Physical review E, 80(5), 056117.

•McDonald, J. H. (2009). Handbook of biological statistics (Vol. 2, pp.173-181). Baltimore, MD: sparky house publishing.

•Morris, S., & Van Der Veer Martens, B. (2008). Mapping research specialties. Annual Review of Information Science and Technology, 42(1), 213–295.

•Peters, H., & Van Raan, A. (1991). Structuring scientific activities by co-author analysis: An expercise on a university faculty level. Scientometrics, 20(1),

235-255.

•Small, H. (1973). Co‐citation in the scientific literature: A new measure of the relationship between two documents. Journal of the American Society

for information Science, 24(4), 265-269.

•Small, H. (1997). Update on science mapping: Creating large document spaces. Scientometrics, 38(2), 275–293.

•Van Eck, N. J., & Waltman, L. (2009). How to normalize cooccurrence data? An analysis of some well‐known similarity measures. Journal of the

American society for information science and technology, 60(8), 1635-1651.

•Zhang, J., Yu, Q., Zheng, F., Long, C., Lu, Z., & Duan, Z. (2016). Comparing keywords plus of WOS and author keywords: A case study of patient

adherence research. Journal of the Association for Information Science and Technology,67(4), 967-972.

Remember to cite:

•Aria, M., & Cuccurullo, C. (2017). bibliometrix:

•An R-tool for comprehensive science mapping analysis.

•Journal of informetrics, 11 (4), 959-975.

Visit our websit to stay updated, to receive tips and suggestions

•https://www.bibliometrix.org/

•info@bibliometrix.org

Support bibliometrix with donations or coding

•https://www.bibliometrix.org/home/index.php/donation/