Analysis of Co-authorship Network and the Correlation between

Academic Performance and Social Network Measures

Qianwen Xu

and Victor Chang

Business Analytics, Xi'an Jiaotong-Liverpool University, Suzhou, China

School of Computing, Engineering and Digital Technologies, Teesside University, Middlesbrough, U.K.

Keywords: Co-authorship Network, Academic Performance, Social Network Analysis, Spearman Correlation Test.

Abstract: This project conducted link analysis and graph cluster analysis to analyze the co-authorship network of 166

researchers, mainly from three top universities in Shanghai, China. The publication data of researchers in the

area of social science between 2014 and 2016 were collected from Scopus, and the g index was calculated as

their performance indicator. For this project, the centrality measures, the efficiency of the egocentric network

were calculated as well as authorities and hubs were identified in the link analysis. In addition, clustering

algorithms based on betweenness centrality were used to conduct the graph cluster analysis. Finally, in order

to identify productive researchers, this project employed the Spearman correlation test to analyze the

correlation between a researcher's performance and social network measures. Results from this test indicate

that except for closeness centrality and degree centrality, the correlation between g-index and betweenness

centrality, eigenvector centrality and efficiency is significant.

1 INTRODUCTION AND

PROJECT AIM

Nowadays, the performance of organizations and

individuals is usually evaluated for the purpose of

management. In the area of academia, researchers are

appraised by assessing their academic performance in

terms of teaching evaluations, research production

and other indicators. Evaluating a researcher's

academic performance is essential as the evaluation

results can be used not only for recruitment and

allocation of funding but also for gaining a high

reputation because of having productive researchers

(Abbasi et al., 2012). However, it is not easy to

identify, cluster and configure productive researchers

to optimize research synergies. In order to address

this question, this project calculated the g index as the

measure of researchers' academic performance and

employing link analysis and graph cluster analysis to

analyze the co-authorship network. Finally, this

project applied a Spearman correlation test to

evaluate the correlation between the researchers'

academic performance and their centrality in the co-

authorship network and the efficiency of their

egocentric network. This paper chooses the Chinese

scholars because although there are numerous studies

on the co-authorship network in China at present, the

papers on the microscopic of the network are little.

The majority of the papers focus on the level of the

nation or a province rather than a city or a

university(Andersson et al., 2014). Fudan University,

Shanghai Jiaotong University and Tongji University

are chosen to be studied in this paper as they are the

top three universities and members of the 985UNIs in

one city, Shanghai. 985UNIs represent the top-level

of the pyramid in China’s higher education system,

their collaboration relationship is close and the data is

sufficient for study(Wang et al., 2014)

2 LITERATURE REVIEW AND

PROPOSED METHODS

2.1 Literature Review

A social network is a set of nodes or actors that are

connected to each other through some kind of

relationship, such as family members, cooperation

between companies and so on. There are usually two

types of social networks, which are socio-centric and

egocentric(Chung et al., 2005). With the development

of social networks, the social network analysis has

Xu, Q. and Chang, V.

Analysis of Co-authorship Network and the Correlation between Academic Performance and Social Network Measures.

DOI: 10.5220/0009428503590366

In Proceedings of the 5th International Conference on Internet of Things, Big Data and Security (IoTBDS 2020), pages 359-366

ISBN: 978-989-758-426-8

359

applied in many topics, such as natural resource

management(Prell et al., 2009), classroom social

interactions(Martı

nez et al., 2003), economic

geography(Ter Wal & Boschma, 2009) and so on. In

the area of academic area, the social network of an

author or a researcher can be an indicator of his co-

authorship with other researchers. Liu et al. (2006)

examine the state of the digital library domain by

analyzing the co-authorship network of the past

ACM, IEEE, and joint ACM/IEEE digital library

conferences. At the same time, Newman learned the

pattern of scientific collaboration from a co-

authorship network (Newman, 2004). In addition,

Morel et al. (2009) found that co-authorship analysis

is a great tool to support the strategic planning of

research on neglected diseases.

2.2 Proposed Methods

In this project, the co-authorship network will be

analyzed by employing link analysis and graph

cluster analysis and a Spearman correlation test will

be conducted to learn the correlation between

academic performance and social network analysis

measures so that a productive researcher can be

identified. From the aspect of link analysis and

Spearman correlation test, the methods this paper will

employ refer to Abbasi et al.'s (2011,2012) in their

research. However, this paper extends their work by

applying the HITS algorithm (Kleinberg, 1998) to

identify the authority and hub of the network. In

addition, a graph cluster analysis based on two types

of betweenness algorithms will be employed. All of

these different analyses and algorithms help us to

make a better understanding of the microscopic of the

co-authorship network in China.

For link analysis, four measures of centrality will be

calculated. According to Freeman(1978), the

centrality of a node impacts leadership, satisfaction

and efficiency significantly. And the performance of

a node is impacted by betweenness centrality and

degree centrality particularly. The centrality

measures calculated in this project are degree

centrality, betweenness centrality, closeness

centrality and eigenvector centrality. The degree of a

node is the number of its adjacent nodes and it is

considered to be the measure of local centrality(Scott,

1991). Betweenness centrality(Borgatti,1995) is

another kind of centrality to measure the degree to

which a given node lies on the shortest paths

(geodesics) between other nodes in the graph.

Closeness(Freeman, 1980) is a measure of a node’s

global centrality by calculating its distance to other

nodes and eigenvector centrality(Bonacich, 1972) is

to measure a node’s centrality based on the concept

that the centrality of a node does not only depend on

the number of its adjacent nodes but also depend on

the centrality of these adjacent nodes.

Based on Burt’s s(Borgatti,1995) structural holes, this

paper also calculated the efficiency of nodes to

evaluate their relationship with authors in one group.

According to Burt, if a node has more primary

contacts from the same group, then the node will

obtain more redundant information from its primary

contacts as nodes within one group usually share the

same information. Therefore, a node’s network is

more efficient if it has a strong relationship with just

one node of a group rather than all authors within the

same group.

Additionally, this project employed Kleinberg’s

(1998) HITS algorithm to identify the authority and

hub of the network. A node is considered as an

authority if it has many pages linking to it and it is

considered as a hub if it points to many other vertices.

After link analysis, this project used two clustering

algorithms based on betweenness centrality to

conduct the graph cluster analysis. The result of the

two algorithms will be compared.

In order to learn how to identify the productive

researchers from their social network measures, the

significance of the relationship between four

centrality measures, efficiency and author's

performance will be evaluated by the Spearman

correlation test(Abbasi et al., 2011). Spearman

correlation test is a tool to evaluate whether two

variables are related to each other

significantly(Gauthier, 2001). The researchers'

performance in this project will be quantified by

using the g index, which was introduced by Egghe

(2006)and widely used by the academic database. The

g index is calculated by ranking a researcher's papers

in decreasing order of their papers’ number of

citations and the g index is the largest number that the

accumulated number of citations the top g papers

received is not less than g2.

The hypothesis tested by Spearman correlation

analysis are as below:

H1: A researcher’s degree centrality impacts his or

her research performance;

H2: A researcher’s betweenness centrality impacts

his or her research performance;

H3: A researcher’s closeness centrality impacts his or

her research performance;

H4: A researcher’s eigenvector centrality impacts his

or her research performance;

H5: A researcher’s efficiency impacts his or her

research performance;

IoTBDS 2020 - 5th International Conference on Internet of Things, Big Data and Security

360

3 IMPLEMENTATION AND

APPLICATION

DEMONSTRATION

3.1 Data Collection and Management

For this project, data of scholars in three top

universities of Shanghai, China, was collected from

the website of Scopus. These three universities are

Fudan University, Shanghai Jiaotong University and

Tongji University. The bibliographic data used in this

project is in the area of social science and about 166

authors’ publication information from 2014 to 2016.

The publication information includes authors’ names,

ID, affiliations, number of publications and number

of citations per paper.

Based on the available published information of

authors, two datasets were built. One is information

about data, including authors' names which are not

full name in consideration of privacy issues, and their

affiliations, number of publications, the total number

of citations by other writers and their g index. The

other dataset contains connections between authors

based on whether there are co-author relationships

between them and the number of cooperation was

assigned to the attribute "weight". Due to the whole

datasets cannot be shown in this paper fully.

Therefore, only a part of the two datasets is selected

randomly and is shown in Table 1 and Table 2.

Table 1: Authors.

Author Affiliation Country

No. of

Publication

Cited

G-

Wang L

Fudan

Universit

China 15 60 7

Li C

Fudan

Universit

China 3 10 3

Zhang

Y-FD

Fudan

Universit

China 4 28 4

Table 2: Co-authorships.

Author 1 Author 2 Weight

Chu N Li D 2

Chu N Gober H -J 2

Chu N Qiu X 2

Harder M K Burford G 2

3.2 Implementation

After preparing the datasets for social network

analysis, this paper used Rstudio and Ucinet (Borgatti

et al., 2002) as tools for visualizing the network and

for calculating the network measures in order to carry

out the later analysis.

First of all, this paper visualized the co-authorship

network through a graph. The nodes of the graph

represent authors. A link between nodes represents a

cooperation relationship between authors and the

width of a link represents the weight of a link,

denoting the number of publications that two authors

cooperated. Fig.1 shows the co-authorship network of

this project.

In this co-authorship network, the green nodes

represent the authors from Shanghai Jiaotong

University, the light blue nodes represent the authors

from Tongji University, the pink nodes represent the

authors from Fudan University while black nodes

were assigned to the authors from other universities.

Before detailed analysis, it can be seen from Fig.1 that

the co-authorship network can be divided into three

groups by university, which is the left upper one, left

lower one and the right side one. And the pink group

which represents the Fudan University, have more

cooperative relationships with organizations outside

the university. Furthermore, the nodes with bigger

sizes seem to play important roles in forming this

network.

Figure 1: Co-authorship network.

Analysis of Co-authorship Network and the Correlation between Academic Performance and Social Network Measures

361

Secondly, link analysis was conducted. Four

centrality measures (normalized degree centrality,

normalized betweenness centrality, normalized

closeness centrality and eigenvector centrality) and

efficiency of every node in structural holes are

calculated. The results are organized and a part of

them are in Table 3 as below:

Table 3: Measures.

Auth

Between

ness

Closen

ess

Degr

Eigenve

ctor

Efficie

ncy

Sun

0 0.0914

0.02

2.22E-07 0.25

Xie J 0 0.0914

0.02

2.22E-07 0.25

Zhu

0 0.0914

0.02

2.22E-07 0.25

Han

0 0.0914

0.02

2.22E-07 0.25

Wei

0 0.0912

0.01

2.01E-07 0.5

Moreover, the HITS algorithm was used to identify

the authority and hub in the co-authorship network.

The algorithm returns two vector columns (hub and

authority) since they are bound together. Therefore,

this paper divided them and a part of them is shown

in Table 4.

Table 4: Authority and Hub.

Author Authority Author Hub

Sun T 1.75E-09 Sun T 1.75E-09

Xie J 1.75E-09 Xie J 1.75E-09

Zhu L 1.75E-09 Zhu L 1.75E-09

Han Z 1.75E-09 Han Z 1.75E-09

Wei WX 1.44E-09 Wei WX 1.44E-09

Chen D 1.72E-08 Chen D 1.72E-08

Xie Y 1.76E-08 Xie Y 1.76E-08

Hu D 1.44E-09 Hu D 1.44E-09

Finally, graph cluster analysis was carried out. An

algorithm based on betweenness centrality was

selected and used for clustering. In the co-authorship

network, it is important for knowledge or academic

information to flow effectively, so identifying a node

or a link that plays the role of a broker is essential.

Therefore, the algorithm based on betweenness

centrality was selected. It helps to identify the vital

nodes or edges. There are two types of betweenness

can be used, which are vertex betweenness and edge

betweenness. Both of them were used in this project

and there is a little difference between the results.

4 ANALYSIS OF RESULTS AND

DISCUSSION

In this section, the results of link analysis will be

analyzed and the Spearman rank correlation test will

be conducted to test the significance of the

relationship between the five measures and the g-

index. Graph cluster analysis will be discussed to

analyze the co-authorship network as well.

4.1 Link Analysis

Among the four measures of centrality, the degree is

the simplest approach of measuring the node

centrality. In this co-authorship network, the average

degree centrality is around 0.033 and Lu H from

Fudan University has the highest degree centrality of

0.091. It means that he or she communicates more

actively than other authors, or he or she is more

prevalent among researchers.

Closeness is a measure of a node’s global centrality

by calculating its distance to other nodes. Among

these authors, their average closeness centrality is

nearly 0.081 and Wang L from Fudan University

gains the highest closeness centrality of 0.135,

meaning that his or her position in this network is the

on average the nearest position to all other authors.

Therefore, he or she is the person who can obtain

information most efficiently.

Betweenness measures the number of times that a

given node lies on the shortest paths between other

nodes in the graph. In this co-authorship network, the

average betweenness centrality is around 0.040 and

Wang L from Fudan University has the highest

betweenness centrality of 0.666. It means that he or

she plays a very important role as a broker or

gatekeeper in the network and he or she can most

frequently control knowledge diffusion among

researchers. If the node of Wang L was missing, then

the single networks of three universities will not be

linked anymore.

Eigenvector Centrality is another measure of a node's

centrality based on the concept that the centrality of a

node does not only depend on the number of its

adjacent nodes but also depends on the centrality of

these adjacent nodes. Among these authors, the

average eigenvector centrality is around 0.33 and

Wang L has the highest value of 1.0. He or She has

nine adjacent authors and more than half of his or her

adjacent authors have high centrality value as well.

From the aspect of the structural hole, efficiency is

the ratio of the total number of disjoint groups of

primary nodes of a node divided by the degree

centrality of the node. In this co-authorship network,

IoTBDS 2020 - 5th International Conference on Internet of Things, Big Data and Security

362

the average efficiency is around 0.36, and the values

of the top 10’s efficiency are greater than 0.9. The

high values indicate that these authors may focus on

a strong relationship with only one co-author of a

group of linked co-authors rather than with all co-

authors within this group and they usually have

access to different kinds of knowledge or academic

information, which will help them to innovate and

perform better than others.

In addition, the lowest value of efficiency is 0.077,

meaning that they tend to have strong relationships

with all co-authors within one group instead of with

one author of this group. In that the same knowledge

will always spread within one group, maintaining a

relationship with all authors of the same group will be

time-wasting as they always get redundant

knowledge from their primary contacts.

From the view of authority and hub, Lu H was

identified to be the authority as well as the hub of the

co-authorship network as it gained the highest value

from both aspects. This means Lu H was considered

to be authoritative and productive in the area of social

science. At the same time, he or she was also linked

to many other researchers who can be considered as

an authority as well. Therefore, Lu H was identified

to be the hub at the same time.

Considering these factors comprehensively, we can

draw a conclusion that comparatively, Wang L seems

to be the most important author in this co-authorship

network, although he was not identified to be the

authority of this network while his or her closeness,

betweenness and eigenvector centrality are the

highest. And as for the degree centrality and

efficiency, his or her value is at the top as well. For

degree centrality, his or her value is 0.055, with an

average value of 0.033 and for efficiency, his or her

value is 0.679, with an average value of 0.36.

Therefore, Wang L plays the most important role in

this network to connect authors and obtain and

transmit knowledge or academic information more

effectively and efficiently.

4.2 Graph Cluster Analysis

After the link analysis, graph cluster analysis was

carried out and two types of betweenness algorithms

were both employed, which are vertex betweenness

and edge betweenness. Fig.2 shows the result of

algorithms based on vertex betweenness and Fig.3

shows the result of algorithms based on edge

betweenness.

The results are similar as both algorithms divided the

co-authorship network into three clusters and most of

the authors from the same university were assigned to

the same cluster.

However, there is still a little difference between the

results. While Wang L is shared by three clusters by

using an algorithm based on vertex betweenness, he

or she belongs to Cluster 1 by using an algorithm

based on edge betweenness. Furthermore, Cluster 1

and Cluster 3 are divided by breaking the link

between Gui Y and Wang J-W in Fig.3 rather than

sharing Wang L, meaning that except the link with

Wang J-W, every nodes or group linked to Gui Y

belong to Cluster 1 rather than Cluster 3 as a result in

Fig.2.

In that, the clusters are divided based on betweenness

centrality and betweenness centrality measures the

number of times that a given node lies on the shortest

paths between other nodes in the graph, the results

indicate that the co-authorship among authors or

researchers from the same university are much closer

than with outside the university although there are

some cases of cooperating with other organizations.

Based on the result of the algorithm of vertex

betweenness, HITs algorithm of link analysis applied

again to identify the authority and hub for each

cluster. And Fan R was identified to be the authority

and hub for the group of Shanghai Jiaotong

University, Yang F for Tongji University and Lu H

for Fudan University.

4.3 Spearman Rank Correlation Test

In order to identify a productive researcher, a

Spearman rank correlation test was conducted to

evaluate whether the correlations between the five

measures and g-index are significant or not and the

result is shown in table 5.

The value to decide whether the correlation is

significant or not was set to be 0.01, meaning that if

the significant value is greater than 0.01, then the

correlation is not significant and if the significant

value is less than 0.01, then the correlation is

significant (Hochberg & Benjamini, 1990). From the

results above, it can be seen that the correlation

between betweenness centrality, eigenvector

centrality, efficiency and g-index is significant as

their significant value is nearly equal to 0 while the

variance in closeness centrality or degree centrality

seems to be not able to explain the variance in g-index

very well as their significant value is far greater than

0.01, which are 0.95 and 0.84 respectively. Therefore,

hypothesis H1, H3 should be rejected and H2, H4 and

H5 should be accepted. According to the coefficient,

it suggests that researchers with higher betweenness

Analysis of Co-authorship Network and the Correlation between Academic Performance and Social Network Measures

363

Figure 2: Co-authorship network – Vertex.Cluster.

Figure 3: Co-authorship network – Edge.Cluster.

Table 5: Spearman rank correlation test.

1 2 3 4 5 6

Betweeness

β 1.00 0.12 0.31** 0.35** 0.39** 0.46**

Sig. . 0.12 0.00 0.00 0.00 0.00

Closeness

β 0.12 1.00 0.11 0.08 0.12 0.01

Sig. 0.12 . 0.16 0.29 0.12 0.95

Degree

β 0.31** 0.11 1.00 0.25** -.63** 0.02

Sig. 0.00 0.16 . 0.00 0.00 0.84

Eigenvector

β 0.35** 0.08 0.25** 1.00 -0.02 0.28**

Sig. 0.00 0.29 0.00 . 0.81 0.00

Efficiency

β 0.39** 0.12 -.63** -0.02 1.00 0.37**

Sig. 0.00 0.12 0.00 0.81 . 0.00

G-I

β 0.46** 0.01 0.02 0.28** 0.37** 1.00

Sig. 0.00 0.95 0.84 0.00 0.00 .

**. Correlation is significant at the 0.01 level (2-tailed).

IoTBDS 2020 - 5th International Conference on Internet of Things, Big Data and Security

364

centrality, higher eigenvector centrality and

efficiency will gain a higher g index, meaning that the

researcher is more productive. The higher

betweenness means that the author or researcher who

is between the cooperation paths, which is between

other authors more frequently and he or she can most

frequently control knowledge diffusion among

researchers. They are more resourceful. The higher

eigenvector centrality indicates that the scholar has

more connections to other authors who are well-

connected as well. The higher efficiency means that

the researcher tends to conduct more collaboration

with diverse researchers from different groups instead

of all authors in the same group.

Based on the result, this paper can conclude that the

author or researcher who collaborates with diverse

authors or groups and or with other authors that are

themselves also well-connected has better academic

performance than those who do not. In addition, the

author or researcher who has strong co-authorship

with just only one author of a group instead of all

authors in the same group perform better than those

who do not.

5 CONCLUSION AND

CONTRIBUTION

5.1 Conclusion

In this project, a co-authorship network of 166

researchers, mainly from three top universities in

Shanghai, China, was analyzed by employing link

analysis and graph cluster analysis. Five social

network analysis measures, degree centrality,

betweenness centrality, closeness centrality,

eigenvector centrality and efficiency were calculated

and the algorithms of HITS and betweenness

clustering were used in the analysis. Results from the

analysis indicate that Wang L is the most important

researcher in this co-authorship network. Finally, in

order to identify productive researchers, this project

employed the Spearman correlation test to analyze the

correlation between a researcher's performance and

social network measures. Results from this test

indicate that except for closeness centrality and

degree centrality, the correlation between g-index and

betweenness centrality, eigenvector centrality and

efficiency is significant.

5.2 Implications

This paper provides the references for the related

institutions and scholars based on the analysis results.

For the related institutions, such as universities or

ministry of education, this paper shows that the co-

authorship network of a scholar is connected to its

academic performance and evaluating the network

may help them to identify, cluster and configure

productive researchers to optimize research

synergies. As for the scholars, this paper may suggest

that scholars should try to collaborate with diverse

authors frequently instead of only one author and

work with authors who are well-connected as well. In

addition, scholars should try to avoid collaborating

with many authors in the same group because this

may lead to low efficiency.

5.3 Contribution and Future Work

This paper gives a glimpse of the internal structure of

the co-authorship network in China. At present, the

majority of the papers that study China's co-

authorship network focus on the level of the nation or

a province rather than a city or a university and the

papers on the microscopic of the co-authorship

network are little. The co-authorship network in this

paper may be small, but it gives the direction of

possible future research. In our future work, the data

of all universities in Shanghai or other cities can be

included to build a complete co-authorship network

at the city level. In addition, more measurements can

be investigated to study their relationship with scholar

performance.

REFERENCES

Abbasi, A., Altmann, J., & Hossain, L. (2011). Identifying

the effects of co-authorship networks on the

performance of scholars: A correlation and regression

analysis of performance measures and social network

analysis measures. Journal of Informetrics, 5(4), 594–

607. https://doi.org/10.1016/j.joi.2011.05.007

Abbasi, A., Chung, K. S. K., & Hossain, L. (2012).

Egocentric analysis of co-authorship network structure,

position and performance. Information Processing &

Management, 48(4), 671–679. https://doi.org/10.1016/

j.ipm.2011.09.001

Andersson, D. E., Gunessee, S., Matthiessen, C. W., &

Find, S. (2014). The Geography of Chinese Science. 22.

Bonacich, P. (1972). Factoring and weighting approaches

to status scores and clique identification. Journal of

Mathematical Sociology, 2(1), 113–120.

Analysis of Co-authorship Network and the Correlation between Academic Performance and Social Network Measures

365

Borgatti, S. P. (1995). Centrality and AIDS. Connections,

18(1), 112–114.

Borgatti, S. P., Everett, M. G., & Freeman, L. C. (2002).

Ucinet for Windows: Software for social network

analysis.

Chung, K. K. S., Hossain, L., & Davis, J. (2005). Exploring

Sociocentric and Egocentric Approaches for Social

Network Analysis. 9.

Egghe, L. (2006). Theory and practise of the g-index.

Scientometrics, 69(1), 131–152. https://doi.org/10.

1007/s11192-006-0144-7

Freeman, L. C. (1978). Centrality in social networks

conceptual clarification. Social Networks, 1(3), 215–

239. https://doi.org/10.1016/0378-8733(78)90021-7

Freeman, L. C. (1980). The gatekeeper, pair-dependency

and structural centrality. Quality and Quantity, 14(4),

585–592.

Gauthier, T. (2001). Detecting Trends Using Spearman’s

Rank Correlation Coefficient. Environmental

Forensics, 2(4), 359–362. https://doi.org/10.1006/enfo.

2001.0061

Hochberg, Y., & Benjamini, Y. (1990). More powerful

procedures for multiple significance testing. Statistics

in Medicine, 9(7), 811–818. https://doi.org/10.1002/

sim.4780090710

Kleinberg, J. M. (1998). Authoritative sources in a

hyperlinked environment. In Proceedings of the ACM-

SIAM Symposium on Discrete Algorithms.

Liu, X., Bollen, J., Nelson, M., & Van, d. S. H. (2005). Co-

authorship networks in the digital library research

community. Information Processing & Management,

41(6), p. 1462-1480.

Martı

nez, A., Dimitriadis, Y., Rubia, B., Gómez, E., & de

la Fuente, P. (2003). Combining qualitative evaluation

and social network analysis for the study of classroom

social interactions. Computers & Education, 41(4), 353–

368. https://doi.org/10.1016/j.compedu.2003.06.001

Morel, C. M., Serruya, S. J., Penna, G. O., & Guimarães, R.

(2009). Co-authorship Network Analysis: A Powerful

Tool for Strategic Planning of Research, Development

and Capacity Building Programs on Neglected

Diseases. PLOS Neglected Tropical Diseases, 3(8),

e501. https://doi.org/10.1371/journal.pntd.0000501

Newman, M. E. J. (2004). Co-authorship networks and

patterns of scientific collaboration. Proceedings of the

National Academy of Sciences, 101(Supplement 1),

5200–5205. https://doi.org/10.1073/pnas.0307545100

Prell, C., Hubacek, K., & Reed, M. (2009). Stakeholder

Analysis and Social Network Analysis in Natural Resource

Management. Society & Natural Resources, 22(6), 501–

518. https://doi.org/10.1080/08941920802199202

Scott, J. (1991). Social Network Analysis: A handbook Sage

Publications London UK.

Ter Wal, A. L. J., & Boschma, R. A. (2009). Applying

social network analysis in economic geography:

Framing some key analytic issues. The Annals of

Regional Science, 43(3), 739–756. https://doi.org/

10.1007/s00168-008-0258-3

Wang, W., Wu, Y., & Pan, Y. (2014). An investigation of

collaborations between top Chinese universities: A new

quantitative approach. 11.

IoTBDS 2020 - 5th International Conference on Internet of Things, Big Data and Security

366