editorial - issi-society.orgissi-society.org/media/1407/newsletter61.pdf · editorial jointly study...

ISSI NEWSLETTER VOL. 16. NR. 1. © International Society for Scientometrics and Informetrics

EDIT

OR

IAL

1ISSI NEWSLETTER VOL. 16. NR. 1.

© International Society for Scientometrics and Informetrics

CONTENTSEDITORIAL

Special Focus: The global

COVID-19 challenge page 1

RESEARCH IN FOCUS

K. Decock, K. Debackere,

A. Vandamme, B. van Looy:

Scenario-driven Forecasting: Lessons Learned from Mod-eling the COVID-19

Pandemic page 2

CONFERENCE REPORT

Conference Report on ISSI2019

page 7

ARTICLES

R. Rousseau: Interdisciplinar-

ity of an Article as Measured through

References: Is Scale-invariance Really

Needed? page 17

P- Chi, Zs. Kosztyán,

L. Gadár, V. Csányi, B. Thijs, A. Telcs,

W. Glänzel: A Multilayer

Network Analysis of Mobility and

Collaboration in European University

Research page 21

1

#61 / Volume 16 number 1March 2020

quaterly e-newsletter of the international society for scientometrics and InformetricsISSN 1998-5460

EDITORIALSPECIAL FOCUS: THE GLOBAL COVID-19 CHALLENGEWOLFGANG GLÄNZEL1 & LIN ZHANG1, 2

1 ECOOM and Dept. MSI, KU Leuven, Belgium2 School of Information Management, Wuhan University, China

This number of the ISSI newsletter comes out with considerable delay. There were and still are severe reasons for that delay: The collection and preparation of the contributions for the present number was accompa-nied and overshadowed by the outbreak and spread of the COVID-19 pandemic. Therefore, we decided to put a relevant article – on modelling the diffusion patterns of COVID-19 and a scenario-driven thinking in forecasting the course of the pandemic – at the heart of the present issue of the Newsletter. The paper by Decock et al. (p. 2–6) is co-authored by our colleagues at KU Leuven. Extended versions and follow-ups of this con-tribution will be published in international scientific journals very soon.

In this context we would like to mention, that in a further article on COVID-19 (Zhang, L., Zhao, WJ., Sun, BB., Huang, Y, Glänzel, W. (2020), How scientific research reacts to international public health emergencies: a global analysis of response patterns, Scientometrics, in press), the two authors of this editorial jointly study with their co-workers how scientific research reacts to international public health emergencies. The article, which is to come in one of the next issues of the journal Scientometrics, provides the bibliometric characteristics and a comparative analysis of global academic response pat-terns to global public-health emergencies in previous pandemics/epidemics of this century with that experienced in the present situation of COVID-19.

ISSI e-Newsletter (ISSN 1998-5460) is published by ISSI (http://www.issi-society.org/).Contributors to the newsletter should contact the editorial board by e-mail.

• Wolfgang Glänzel, Editor-in-Chief: wolfgang.glanzel[at]kuleuven.be • Balázs Schlemmer, Managing Editor: balazs.schlemmer[at]gmail.com• Sarah Heeffer, Assistant Editor: sarah.heeffer[at]kuleuven.be• Judit Bar-Ilan • Sujit Bhattacharya: sujit_academic[at]yahoo.com • María Bordons: mbordons[at]cchs.csic.es • Juan Gorraiz: juan.gorraiz[at]univie.ac.at • Jacqueline Leta: jleta[at]bioqmed.ufrj.br • Olle Persson: olle.persson[at]soc.umu.se • Ronald Rousseau: ronald.rousseau[at]kuleuven.be • Dietmar Wolfram: dwolfram[at]uwm.edu

Accepted contributions are moderated by the board. Guidelines for contributors can be found at http://www.issi-society.org/editorial.html. Opinions expressed by contributors to the Newsletter do not necessarily reflect the official position of ISSI. Although all published mate-rial is expected to conform to ethical standards, no responsibility is assumed by ISSI and the Editorial Board for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material therein.

http://www.issi-society.org/

http://www.issi-society.org/editorial.html

ISSI NEWSLETTER VOL. 16. NR. 1. © International Society for Scientometrics and Informetrics 2

RES

EAR

CH IN

FO

CUS:

CO

VID

-19 P

AN

DEM

IC SCENARIO-DRIVEN FORECASTING: LESSONS LEARNED FROM MODELING THE COVID-19 PANDEMIC

1 Department of Management, Strategy & Innovation, Faculty of Economics & Business, KU Leuven, Belgium

2 Flanders Business School, Belgium3 KU Leuven, Department of Microbiology, Immunology and Transplantation,

Rega Institute for Medical Research, Clinical and Epidemiological Virology, Institute for the Future, Leuven, Belgium;

4 Center for Global Health and Tropical Medicine, Unidade de Microbiologia, Instituto de Higiene e Medicina Tropical, Universidade Nova de Lisboa, Lisbon, Portugal

KRISTOF DECOCK 1, 2

KOENRAAD DEBACKERE 1

ANNE-MIEKE VANDAMME 3, 4

BART VAN LOOY 1, 2

SUMMARY: The Forecasting patterns and their implied end states remains cumbersome when few (stochastic) data points are available during the early stage of diffusion processes. Extrapolations based on compounded growth rates do not account for inflection points nor end-states. In order to remedy this situation, we advance a set of heuristics that blend forecasting and scenario thinking. Combined they provide an actionable decision space as short term predictions are accurate, while a portfolio of different end states informs the long view.

The creation of such a decision space requires temporal distance; only to the extent that one re-frains from incorporating more recent data, more plausible end states become visible. As such, our contribution implies a plea for dynamically blending forecasting algorithms and scenario oriented thinking, rather than conceiving them as substitutes or complements.

ACKNOWLEDGMENT: This contribution benefited from useful input and reflections by Jorge Ri-cardo Blanco Nova (KU Leuven), Michela Bergamini (KU Leuven), Sien Luyten (Flanders Business School) and Xiaoyan Song (KU Leuven).


RES

EAR

CH IN

FO

CUS:

CO

VID

-19 P

AN

DEM

IC INTRODUCTION: MODELING DIFFUSION PATTERNS OF COVID-19 PANDEMIC

The current COVID-19 pandemic spurred efforts to model and forecast its diffusion patterns, either in terms of infections, peo-ple in need of medical assistance (hospitali-zation, ICU occupation) or casualties.

Forecasting the evolution of COVID-19 and related dynamics (e.g. ICU occupation) remains cumbersome when only few sto-chastic data points are available during the early stage of the outbreak. Extrapolation based on compounded growth rates could result in unstable predictions as they do not include inflection points (i.e. the peak of the net ICU increase) nor end states (i.e. the maximum ICU capacity). Several forecast-ing models suggest that the implied growth patterns follow sigmoid growth curves, like the logistic equation (Verhulst, 1838) or the Bass (1969) model. Predicting the evolution of these S-shaped curves involves an estima-tion of at least three parameters, related to the takeoff and the steepness of the curve as well as its saturation level. However, delin-eating robust parameter estimations – and consequently portraying the end-state – based on a limited time series results in quite rudimentary predictions. Recently, Decock, Debackere & Van Looy (2020) adapted the initial Bass (1969) model in order to quan-titatively model different diffusion scenar-ios for Electrical Vehicles. The heuristics advanced start from the premise of multi-finality (Buckley, 1967): “similar initial con-ditions may lead to dis-similar end-states”. The proposed heuristics revolve around the development of a three-dimensional search space, reflecting the presence of three mod-el parameters to be estimated.

In the context of the COVID-19 pan-demic, it remains unclear at this stage which end states will occur with respect to e.g. casualties, capacity of ICU beds need-ed at its highest burden (in different re-gions/countries) as the available numbers for a wide range of countries suggest high

spreads (both in terms of contamination and potential end states). This complicates the quest for accurate predictions, unless one considers the potential presence of multiple end states and different growth dynamics (reflected in the three param-eters to gauge) simultaneously.

As such, the COVID-19 diffusion pat-terns for Belgium have been modeled by means of the heuristics proposed by Decock, Debackere and Van Looy to see whether they allow to arrive at a better in-formed decision space, both in the short (daily/weekly) and in the medium term (end states). Models have been developed both for deceased and for ICU occupation rates. In this contribution, we focus on the ICU models, due to space constraints (the insights are similar across indicators).

The considered search grid consists of 250,000 different parameter combinations (consisting of 10 different end-states sce-narios multiplied with 250 values for the relevant infection parameter and 100 values for the relevant contamination parameter). By means of a loss function, all 250,000 pathways are assessed in terms of how well they explain the already available observa-tions. In this exercise, the diffusion paths with a goodness-of-fit exceeding 99 percent have been considered as more likely scenar-ios and have been further analyzed in terms of growth dynamics and end states.

ICU OCCUPATION IN BELGIUM

In Belgium, a first COVID-19 deceased was registered on March 10. With an initial overall capacity around 1,900 ICU beds in the Belgian hospitals, policy makers decid-ed to increase ICU capacity and allocated approximately 2,300 ICU beds exclusively for COVID-19 patients. The question that then becomes crucial: will this be enough for the coming weeks and months?

Table 1 depicts the number of patients in ICU (related to COVID-19) in Belgium, between March 12 and March 24:


RES

EAR

CH IN

FO

CUS:

CO

VID

-19 P

AN

DEM

IC

Building on the observations until March 24 (i.e. time series of 13 data points), a multi-dimensional search space was assessed im-plying a potentially required ICU-capacity

ranging from 1,000 to 10,000 beds. Figure 1 visualizes the initial observations (i.e., the green dotted line) combined with the stylized more likely scenarios (averaged by end-state).

Table 1 COVID-19 ICU occupation in Belgium between 12/03/2020 and 24/03/2020 (Source: Sciensano)

DATE ICU OCCUPATION DATE ICU

OCCUPATION DATE ICU OCCUPATION

12/03/20 5 17/03/20 100 22/03/20 322

13/03/20 24 18/03/20 130 23/03/20 381

14/03/20 33 19/03/20 164 24/03/20 474

15/03/20 53 20/03/20 238 16/03/20 79 21/03/20 290

Table 2 COVID-19 ICU occupation in Belgium between 25/03/2020 and 04/04/2020 (Source: Sciensano)

DATE ICU OCCUPATION DATE ICU

OCCUPATION DATE ICU OCCUPATION

25/03/20 605 29/03/20 927 02/04/20 1,205

26/03/20 690 30/03/20 1,021 03/04/20 1,245

27/03/20 789 31/03/20 1,088 04/04/20 1.261

28/03/20 867 01/04/20 1,144

7000

6000

5000

4000

3000

2000

1000

0

12 M

arch

14 M

arch

16 M

arch

18 M

arch

20 M

arch

22 M

arch

24 M

arch

26 M

arch

28 M

arch

30 M

arch

01 A

pril

03 A

pril

05 A

pril

07 A

pril

09 A

pril

11 A

pril

13 A

pril

15 A

pril

ICU

OCC

UPA

TIO

N IN

BEL

GIU

M

1 000 2 000 3 0004 000 5 000 60007 000 8 000 9 00010 000Observations used for model compositionObservations used for model validation

Fig 1 Overview of stylized pathways reflecting more likely scenarios, including the observations used for model composition (i.e. green dotted line) and the unfolding observations afterwards (i.e. red dotted line). Belgium, R2 > .99; n = 806.


RES

EAR

CH IN

FO

CUS:

CO

VID

-19 P

AN

DEM

IC 160

140

120

100

80

60

40

20

0

14 M

arch

16 M

arch

18 M

arch

20 M

arch

22 M

arch

24 M

arch

26 M

arch

28 M

arch

30 M

arch

01 A

pril

03 A

pril

05 A

pril

07 A

pril

09 A

pril

11 A

pril

13 A

pril

15 A

pril

17 A

pril

19 A

pril

21 A

pril

23 A

pril

25 A

pril

27 A

pril

29 A

pril

NET

INCR

EASE

IN IC

U O

CCU

PATI

ON

160

140

120

100

80

60

40

20

0

15 M

arch

17 M

arch

19 M

arch

21 M

arch

23 M

arch

25 M

arch

27 M

arch

29 M

arch

31 M

arch

02 A

pril

04 A

pril

06 A

pril

08 A

pril

10 A

pril

12 A

pril

14 A

pril

16 A

pril

18 A

pril

20 A

pril

22 A

pril

24 A

pril

26 A

pril

28 A

pril

30 A

pril

NET

INCR

EASE

IN IC

U O

CCU

PATI

ON

Fig 2 Comparing the daily net increase in ICU occupation (i.e., dotted line) with the “more likely” scenarios (R2 > .99) in Belgium.

1200 27003000 Observations3300 3600 3900

1500 1800 2100 2400

Observations1000 2000 3000


RES

EAR

CH IN

FO

CUS:

CO

VID

-19 P

AN

DEM

IC As the data of the 11 days between March 25 and April 4 became available (see table 2 and visualized as red dots in figure 1 below), it becomes feasible to further qualify the likelihood of different pathways and their corresponding end-states.

As from March 28 onwards, it became clear that the end state of ICU occupation in Belgium was more likely to become situ-ated in a range between 1,000 and 3,000. Consecutive updates and refinements in terms of feasible end states allowed us to predict the growth pattern and the peaks even more precisely, as depicted in fig-ure 2. On April 2nd we informed the Bel-gian authorities that the growth peak has been reached and that the coming days, the growth rate for ICU capacity would decline markedly. This prediction materialized as shown by the red dots in figure 2.

DISCUSSION AND CONCLUSION

In this contribution we illustrate how blend-ing forecasting models with scenario-ori-ented thinking yields novel insights, which inform decision makers in a number of dis-tinctive ways. The heuristics implied, com-bine a broad range of end states with an assessment of more likely pathways. Not only do these models provide accurate pre-dictions in the short term; when additional observations become available, they also signal plausible end-states in the scenario portfolio. The creation of such a decision space requires temporal distance: only to the extent that one refrains from incorporating immediately more recent data, more plau-sible end states become visible. At the same time, updating and re-calibrating the path-ways seem to offer potential to start qualify-ing Knightian uncertainty (Knight, 1921).

As such, our contribution implies a plea for combining forecasting algorithms with a foresight-oriented methodology and vice versa. This dynamic blending of forecast and foresight has the potential to inform policy makers in situations of urgent deci-sion needs conditioned by profound un-certainty. At such critical moments dur-ing an unfolding crisis, the use of large amounts of data cannot inform decision-making, as those data are largely absent in such instances. Though, it is the judicious use of the limited data available combined with various scenarios that may unfold. Our combination of forecast and foresight hence does not signal yet another case in evidence-based policy, but it rather illus-trates the good governance of evidence as advocated by Parkhurst (2016).

REFERENCES

Bass, F.M. (1969). A New Product Growth for Model Consumer Durables. Management Science, 15 (5), 215-227.

Buckley, W. (1967) Sociology and Modern Systems Theory. Prentice-Hall, Englewood Cliffs.

Decock, K., Debackere, K. and Van Looy, B. (2020). Bass Re-visited: Quantifying Multi-Finality. MSI working paper, n° 2003.

Knight F. (1921) Risk, uncertainty and profit. Houghton Mifflin Company, Boston/New York.

Parkhurst, J. (2016) The Politics of Evidence: From Evidence-based Policy to the good Governance of Evidence, Routledge.

Verhulst, P.F. (1838). Notice sur la loi que la popula-tion suit dans son accroissement. Correspon-dance mathématique et physique, 10, 113–121.


CON

FER

ENCE

REP

OR

T

7

CONFERENCE REPORT ON ISSI2019

CINZIA DARAIOSapienza University of Rome, Italy

The 17th International Conference of the International Society for Scientometrics and Informetrics (ISSI2019) took place at Sa-pienza University of Rome on 2-5 September 2019. For the first time, Italy hosted an ISSI Conference and the city of Rome, the “eter-nal city”, offered a beautiful frame to the Conference. As we wrote in the last email to all conference participants: “we waited for our conference for years (it was discussed and approved by the ISSI Board in 2015 and confirmed in 2017) but it passed like a flash, remaining in our mind as a dream...” Any-way, we have the gadgets of the conference to remind us that it happened…

The conference has been a major event attended by more than 650 participants from 44 countries from all the world. The conference included a special STI Indica-tors Conference Track organized in col-laboration with the European Network of Indicator Designers (ENID).

The program of the four days was very dense, although we tried to organize the par-allel, the plenary and the posters sessions, in such a way to balance the different themes at the benefit of a variegated audience.The first day of the conference, on Mon-day 2 September 2019 we had several

interesting workshops and sponsored ses-sions, ranging from Altmetrics and open citations, to impact assessment and appli-cation of Big Data in Scientometrics, in-cluding dedicated platforms (CADRE and NETSCITY), collective datasets (ETER), in-

The conference bag from Made in carcere (producer of ethical gadgets made in jail by women from the South of Italy) and Sapienza’s gadgets for remem-bering the time spent in Rome during ISSI2019.Photo courtesy of © Cinzia Daraio


CON

FER

ENCE

REP

OR

T

8

vestigation on scholarly communication in China and the study of review articles. The program of the first conference day was completed by a rich Doctoral Forum, and a light welcome dinner preceded the Jazz concert performed by Musa Jazz, one of the Sapienza’s Orchestras, in the Aula Magna.

All posters were categorized using the same classification into themes as that used to group the papers. Next, they were aggregated into four main poster groups: i) Indicators & databases, ii) Social context, innovation, policy, iii) Analytical methods and iv) Application domains.

The poster session around the poster boards was preceded by a session in which

all posters in a particular group were pre-sented in short presentations of about one minute, in a conference room.

The Conference Proceedings of more than 2800 pages organized in two volumes contain all the contributions presented at the conference: around 260 papers and more than 150 posters.

All the topics presented during the plena-ry sessions, although addressed by scholars with different backgrounds and analysing different aspects, were around the multidi-mensionality and the complexity of the assess-ment in Scientometrics and Informetrics.

The first keynote was delivered by Prof. John Carson from the University of Michigan

Concert of Musa Jazz Orchestra, Monday 2 September 2019 in Aula Magna. Photo: © Maurizio Perciballi

Table 1 All accepted papers were grouped into the following eighteen themes and were grouped according to their theme into eight parallel lines:

Altmetrics Higher Education Systems Patent analysis

Automated Text Processing Humanities and Social Sciences Research evaluation models

Databases Informetric distributions Research field mapping

Funding and determinants Library science aspects Scholarly publishing

Gender studies National R&D systems Social context of R&D

Globalization & collaboration Novel bibliometric indicators Technological innovation


CON

FER

ENCE

REP

OR

T

9

(USA), entitled “Quantification, its affordanc-es and limits”. John Carson is an expert in the history of quantification and large part of his research has been devoted to the quantifica-

tion and measurement of intelligence. We recommend his very interesting, deep and inspiring book on “The Measure of Merit: Talents, Intelligence, and Inequality in the

Conference Proceedings. Photo: © Maurizio Perciballi

First keynote, Tuesday 3 September 2019, John Carson, University of Michigan (USA) and his book “The Meas-ure of Merit: Talents, Intelligence, and Inequality in the French and American Republics, 1750-1940”Photo: © Maurizio Perciballi


CON

FER

ENCE

REP

OR

T

10

French and American Republics, 1750-1940”. In his plenary talk, Carson pointed out some crucial aspects connected to the evaluation that every scholar in our field should consid-er in our opinion. He started his talk by con-sidering quantification as “moral endeavor”.

Quantification can never stand alone, Carson said, it is as much human as it is me-chanical, subjective as well as objective. “Why quantify? The world is messy and complex and full of noise; combining bits of informa-tion and analyzing them rigorously is often the only way to detect patterns and truths concealed in the flood of information with which we are inundated. Every act of seeing, however, is also an act of occlusion, of not-seeing. To attend to one part of the visual spectrum one must pay much less attention to other parts; to count one kind of thing, one cannot include other kinds.” Quantifica-tion is not just a technical pursuit, but a moral one; the reductions and simplifications it re-lies on can bring great and important clarity, but always at a cost. The moral questions for the practitioner are first, is the cost justified, and second, and even more critically, who will pay it? The powerful or the powerless? The established or those fighting for a place? After

a description of the problems of archives in history that are almost the same as those that apply to data selection and treatment in our filed, John Carson called for “epistemic modes-ty”, that is for scholars in all fields to be aware of the limits of their work, its partialness, and the particular standpoint from which they are coming. By applying his suggestion, we should then try to (1) take care to make the framing of problems to be explored expansive and inclusive; (2) pay particular attention to the most vulnerable; (3) insure that our work is distributed to a wide network of those who might be affected by it so that they have a chance to respond; and (4) consider whatever we come up with to be provisional and open to revision as we learn more. “None of these techniques can ensure that we don’t make mistakes, but they should at least remind us to think hard about the silences our archive will contain and to think carefully about who might bear the costs of whatever we have ren-dered invisible” he finally concluded.

The second keynote was given by Prof. Mike Thelwall, from the University of Wol-verhampton (UK), very well known in our field and recipient of the prestigious Derek de Solla Price Award, with the title “New

Second keynote, Wednesday 4 September 2019, Mike Thelwall, University of Wolverhampton (UK) and his book “Web Indicators for Research Evaluation: A Practical Guide”. Photo: © Maurizio Perciballi


CON

FER

ENCE

REP

OR

T

11

developments in scientometric and in-formetric research”. In his talk, Thelwall discussed three recent developments that are changing and challenging scientomet-ric practice: altmetrics, full text mining, and impact assessment. After decades of research into altmetrics and webomet-rics, alternative indicators have emerged as a standard part of scholarly communi-cation infrastructure. This can be seen in the availability of Altmetric.com scores in many publisher websites and the infor-mal use of altmetrics supporting research evaluation narratives. Altmetrics are well enough understood that we can now rec-ommend appropriate uses, and are ready to fully exploit them. Of particular interest for the readers is Mike Thelwall’s nice book “Web Indicators for Research Evaluation: A Practical Guide”. This book describes how to analyse altmetric and webometric data for research evaluation purposes, including appropriate methods for field normalising it and reporting it for use in evaluations.

Full text mining is a second major devel-opment presented by Mike Thelwall, which is taking advantage of the increasing avail-ability of collections of open access docu-ments, such as the PubMed Central Open Access Subset, to get fine-grained informa-tion about citation contexts. This approach has the potential to identify important types of citation to get more precise evidence of the type of impact reflected by citation counts. The impact agenda in the UK Re-search Excellence Framework (REF) was the object of the last part of Thelwall’s talk and was presented as the requirement for re-searchers to produce evidence-based narra-tives describing how their research has non-academic societal benefits. This has led to changes within UK universities to promote societal impact, such as through the ap-pointment of impact officers. Altmetric and Webometric indicators are used, Thelwall said, as part of the evidence base of some of these narratives, which provide a tough chal-lenge for scientometricians. His talk ended with a few considerations in line with John

Carson’s talk, with a few words about the im-portance of using research indicators respon-sibly. Practitioners should always be aware of potential biasing and systemic effects that can lead to unintended consequences.

The plenary sessions were enriched by the celebration of the Eugene Garfield Award, the celebration of the Derek de Solla Price Award and that of the Best ISSI Paper Award. In addition, within the plenary ses-sions, five invited lectures were presented. These lectures were selected among the sub-mitted papers at the conference highly ap-preciated by the reviewers, and that in the view of the Program Chairs transcend the boundaries of their specialized topic, by put-ting this topic in a broader perspective, and/or by combining methods or findings from different specialized domains. The list of the invited lectures is reported in the followings:

► Invited lecture 1: Frederique Bone and Daniele Rotolo. Text-Mining Histori-cal Sources to Trace Technical Change: The Case of Mass Production

► Invited lecture 2: Grit Laudel. Studying the embeddedness of researchers’ ca-reers: Can bibliometric methods help?

► Invited lecture 3: Tao Han and Xiaoyu Cai. Analysis of Division of Labor in High Quality Life Science Research of China

► Invited lecture 4: Francois Van Schalk-wyk, Jonathan Dudek and Rodrigo Costas. Using altmetrics to study social movements and cognitive bridges in the communication of science in the social media: The case of the anti-vacci-nation movement on Twitter

► Invited lecture 5: Matthias Held and Theresa Velden. How to interpret algo-rithmically constructed topical struc-tures of research specialties? A case study comparing an internal and an external mapping of the topical struc-ture of invasion biology


CON

FER

ENCE

REP

OR

T

12

Palazzo Doria Pamphilj: welcome cocktail, mirrors hall and cloister just before the conference dinner, Wednes-day 4 September 2019. Photo: © Maurizio Perciballi


CON

FER

ENCE

REP

OR

T

13

The Conference Dinner was held at Palazzo Doria Pamphilj, at the center of the cent-er of Rome, on Wednesday 4 September, 2019. Some of the photos reported here were taken during the Welcome Cocktail, in the cloister where the dinner was held,

after the guided tours of the art galleries, including the extraordinary hall of mirrors. We take the opportunity here to thank very much all our Sponsors including Elsevier, Digital Science, Data for Impact, Clari-vate Analytics, RISIS, Frontiers, Journal of

Special Plenary session “Memorial of Judit Bar-Illan” Thursday 5 September 2019. Photo: © Maurizio Perciballi


CON

FER

ENCE

REP

OR

T

14

Data and Information Science and Quan-titative Science Studies, for their generous contributions that allowed us to organize the conference and all its social events in a comfortable way.

The special plenary session organized by Gali Halevi on the 5 September 2019 and devoted to the Memorial of Judit Bar-Ilan, who passed away on July 16, 2019, was par-ticularly touching.

Another special plenary session entitled “Evaluative informetrics– the art of metrics-based research assessment”, including the conferral of a Doctorate Honoris Causa in Research Assessment Methodologies award-ed by the Rector of Sapienza University of Rome to Prof. Henk F. Moed took place in the second part of the plenary session of the 5 September 2019. During this session, Prof. Wolfgang Glänzel introduced the content of the forthcoming book by Springer Na-ture on “Evaluative informetrics: the art of metrics-based research assessment – Fest-schrift in honour of Henk F. Moed” edited by Cinzia Daraio and Wolfgang Glänzel and including several original contributions by well-known scholars.

In the last part of this special session, Henk Moed delivered his Lectio Magistralis

Special plenary session on “Evaluative informetrics– the art of metrics-based research assessment” organ-ized in honour of Henk Moed Thursday 5 September 2019 and his book “Applied Evaluative Informetrics”. Photo: © Maurizio Perciballi

on “The Application Context of Research Assessment Methodologies”, just after the presentation of the paper by Yong Zhao, Jiayan Han, Jian Du and Yishan Wu on: “Origin and Impact: A Study on the Intel-lectual Transfer of Professor Henk Moed”. In his Lectio Magistralis, Moed discussed a complex and controversial topic that is the use of bibliometric or informetric indica-tors in the assessment of research perfor-mance. His aims were to spread light on the application context of these indica-tors. He proposed an application model for the use of indicators in the assessment of academic research, highlighting a series of analytical distinctions and useful building blocks, to contribute in this way to enlight-ening and further developing the search for best practices in research assessment. His proposals was organized in three sec-tions. After an introduction on the use of indicators in research assessment, based on his recent monograph “Applied Evalua-tive Informetrics” published by the author in 2017 that we highly recommend to the readers, he proposed a series of useful ana-lytical distinctions. These analytical dis-tinctions were useful to draw conclusions on the validity and usefulness of the most


CON

FER

ENCE

REP

OR

T

15

common indicators of the performance of three principal units of assessment: indi-vidual researchers, research institutions and scientific-scholarly journals. Finally, an application model for the use of indica-tors in academic research was presented, and an extended version can be found by the readers in his new journal Scholarly Assessment Reports (Moed, H.F., 2020. Ap-propriate Use of Metrics in Research As-sessment of Autonomous Academic Insti-tutions. Scholarly Assessment Reports, 2(1),

Final greetings at the end of the Closing session of the conference, Thursday 5 September 2019. Photo: © Maurizio Perciballi

p.1. DOI: http://doi.org/10.29024/sar.8) open now for submissions at https://www.schol-arlyassessmentreports.org/

Finally, we like to end this conference report with a photo and the lightness and the view towards the blue sky of a song. The photo, reported below, was taken at the end of the closing session of the conference, and the song is “Volare (to fly)” by Domenico Mo-dugno, a classical Italian song played during the concert of the Musa Jazz of Sapienza Or-chestra the first day of the conference.

http://doi.org/10.29024/sar.8

https://www.scholarlyassessmentreports.org/

https://www.scholarlyassessmentreports.org/


CON

FER

ENCE

REP

OR

T

16

Penso che un sogno così non ritorni mai più mi dipingevo le mani e la faccia di blu poi d'improvviso venivo dal vento rapito e incominciavo a volare nel cielo infinito

Volare oh, oh cantare oh, oh nel blu dipinto di blu felice di stare lassù e volavo, volavo felice più in alto del sole ed ancora più su mentre il mondo pian piano spariva lontano laggiù una musica dolce suonava soltanto per me

Volare oh, oh cantare oh, oh nel blu dipinto di blu felice di stare lassù ma tutti i sogni nell'alba svaniscon perché quando tramonta la luna li porta con sé ma io continuo a sognare negli occhi tuoi belli che sono blu come un cielo trapunto di stelle

Volare oh, oh cantare oh, oh nel blu degli occhi tuoi blu felice di stare quaggiù e continuo a volare felice più in alto del sole ed ancora più su mentre il mondo pian piano scompare negli occhi tuoi blu la tua voce è una musica dolce che suona per me

Volare oh, oh cantare oh, oh nel blu degli occhi tuoi blu felice di stare quaggiù nel blu degli occhi tuoi blu felice di stare quaggiù con te.

I think that dream like that will never return I painted my hands and face blue Then all of sudden I was abducted by the wind And I started to fly in the infinite sky

To fly ho ho To sing ho ho hoho In the blu, painted blue Happy to stay up there, And I flew and flew happy, higher than the sun And yet more up, While the world slowly disappeared down below, A sweet music playing just for me

To fly ho ho To sing ho ho hoho In the blu, painted blue Happy to stay up there But all dreams vanish at dawn because When the moon sets, it takes dreams with itself But I keep dreaming in your beautiful eyes, That are blue as the sky spangled with stars

To fly, ho ho To sing ho ho hoho, In the blue of your blue eyes Happy to be down here And I keep flying, happy happy, higher than the sun and yet more up, While the world is slowly vanishing in your blue eyes Your voice is a soft music that plays for me

To fly ho ho To sing ho ho hoho In the blue of your blue eyes Happy to be down here In the blue of your blue eyes Happy to be down here With you.

(English translation: https://lyricstranslate.com )

“Volare”by Domenico Modugno

https://lyricstranslate.com/it/nel-blu-dipinto-di-blu-blue-painted-blue.html


SHO

RT

COM

MU

NIC

ATI

ON

S, A

RTI

CLES

17

INTERDISCIPLINARITY OF AN ARTICLE AS MEASURED THROUGH REFERENCES: IS SCALE-INVARIANCE REALLY NEEDED?

RONALD ROUSSEAUUniversity of Antwerp, Faculty of Social Sciences, Antwerpen, [email protected] for R&D Monitoring (ECOOM) and Dept. MSI, KU Leuven, Leuven, [email protected]

1. INTRODUCTION: INTERDISCIPLINARITY

When considering the term interdisciplinar-ity I first like to point out that defining a con-cept and measuring it are two different things. Moreover, the notion of interdisciplinarity includes another one, namely that of a disci-pline. I think it is safe to say that the notion of a discipline cannot be formulated in a way that covers all possible situations and applica-tions. In practice, the notion of a discipline or

knowledge domain is operationalized in three broad ways (Zitt et al., 2019): using ready-made classifications, using an algorithmic approach and determining the knowledge domain of an article through the domains to which its authors belong. In this note, I will not go deeper into this aspect and assume that an appropriate choice has been made.

Next, I focus on one way to determine the interdisciplinarity of an article, namely through the interdisciplinarity of its refer-

ABSTRACT: I argue that scale-invariance may not be a necessary condition for measuring interdisci-plinarity through the diversity of references.

Keywords: scale-invariance; Shorrocks’ generalized Lorenz curve; references; interdisciplinarity


SHO

RT

COM

MU

NIC

ATI

ON

S, A

RTI

CLES

18

ences, or of the set of references expanded with the set of references of references, i.e. us-ing one or two backward generations (Rafols & Meyer, 2010; Hu et al., 2011). Further on in this note I will use the term references when referring to one or more generations of refer-ences. For the next step, there are again many alternatives, but I focus on one often applied method, namely applying a diversity meas-ure on the set of disciplines covered/ used in these references.

This note is further subdivided as follows: first I give a short introduction about how interdisciplinarity can be measured. Next, I come to the main part of this note, namely introducing the question if scale-invariance is really a requirement when studying inter-disciplinarity through the diversity of used references. One suggestion is offered to measure diversity in a way that is not scale-invariant. Finally, a short list is offered of problems that must be solved if one wants to continue in the direction of non-scale-invar-iance for studying interdisciplinarity.

2. MEASURING DIVERSITY

Traditionally diversity consists of two con-cepts: variety and balance. Variety refers to the number of cells, i.e., the number of nominal classes such as species in biology and disciplines in informetric studies, while evenness or balance is defined as the relative apportionment of abundances among cells, actually present, or assumed to be possibly present (Rousseau & Van Hecke, 1999). It has been shown that evenness on its own is best represented by a classical Lorenz curve (Lorenz, 1905; Nijssen et al., 1998). Yet, it has been convincingly argued that, besides vari-ety and evenness, a third notion must be tak-en into account when measuring diversity, namely the disparity (or its opposite: prox-imity) between cells. Stirling (2007) points out that species in ecology or WoS catego-ries in informetrics are not independent entities, but they are shaped by patterns of common developments or ancestry leading to proximities between units, see also (Lein-

ster & Cobbold, 2012; Zhang et al., 2016). Fi-nally, Jost (2006, 2009) pointed out that one should use measures that make it possible to discuss percentage changes in a meaning-ful way, see also (Leinster & Cobbold, 2012; Zhang et al., 2016). For a review on inter-disciplinarity I refer to (Wagner et al., 2011; Rousseau et al., 2019).

Next, I introduce some notation. Let X = [x1, …, xN] be an array of N sources or cells (N a natural number larger than 1), ranked according to the number of items (a non-negative value) in each cell. In this note, I will refer to such arrays as diversity arrays. Let f be a diversity measure and f(X) the value of f for the array X. This measure f is said to be scale-invariant if, for each c > 0, f(cX) = f(X). In most cases being scale-invariant is consid-ered a necessary property for diversity (and concentration) measures. The idea behind this requirement is that diversity or con-centration should not depend on monetary units (dollars, euros, Chinese yuan) in the case of income studies, or weight units (kg, g, pound) in case of biomass. All measures based on the Lorenz curve, such as the Gini index or the coefficient of variation, are by definition scale-invariant as the Lorenz curve is constructed using relative values (see fur-ther), not absolute ones.

3. IS SCALE-INVARIANCE REALLY NEEDED?

I return now to the specific case I want to study, namely measuring interdisciplinarity of an article through the diversity of its refer-ences. If article A1 is based on four disciplines occurring 10, 6, 2 and 1 times, while article A2 is also based on the same four disciplines occurring 20, 12, 4 and 2 times, which one reflects the most interdisciplinarity? In this hypothetical case, I assume that there are no spurious references and no redundancies: each reference is ‘really’ needed. Then classi-cal approaches consider these two articles to be equally interdisciplinary.

Yet, because each article, theoretically, makes new and different knowledge claims, it is clear that the second article used the


SHO

RT

COM

MU

NIC

ATI

ON

S, A

RTI

CLES

19

knowledge embedded in these four disci-plines more thoroughly than article A1. For this reason, one may say that A2 bears tes-timony of more interdisciplinarity than A1. Consequently, I claim that depending on the purpose of the investigation, scale-invari-ance is not a necessary property.

4. HOW TO REPRESENT “NON-SCALE-INVARIANCE”?

How can the above observation be taken into account? Already in 1983 Shorrocks when studying income inequality, observed that, if for a fixed number of cells, the par-tial order introduced by the Lorenz curve, which is actually the dominance order as studied e.g. in (Hardy et al., 1934), is often in-conclusive among two situations, as Lorenz curves tend to cross and for this reason, are intrinsically incomparable (Shorrocks, 1983). For this reason, he proposed the generalized Lorenz curve, constructed by scaling up the ordinary Lorenz curve by the mean of the distribution. This construction reduced the number of crossing curves.

I next recall the construction of the ordi-nary Lorenz curve (Lorenz, 1905; Rousseau et al., 2018) and then build on this to de-scribe Shorrocks’ generalized Lorenz curve. Given the diversity array X = [x1, …, xN], the ordinary Lorenz curve (Lorenz, 1905) is constructed as follows: for j = 1, …, N, let

1

jj ji

S x=

=∑be the j-th partial sum and hence sN = TOT the total number of items. Moreover, s0 is set equal to 0. Now, take the points with coordinates

, kskN TOT

,

k = 0, …, N and connect them by line seg-ments to obtain a curve joining the origin (0,0) with the point (1,1). This is the ordi-nary Lorenz curve. If x-values are equal the Lorenz curve is a straight line (the diagonal of the unit square). If the array X is ordered decreasingly, the Lorenz curve is concave and situated above this straight line; if X is

ordered increasingly, as done by Shorrocks, then the Lorenz curve is convex and situ-ated under the diagonal.

Now for Shorrocks’ generalized Lorenz curve, the abscissa values are the same as for the ordinary Lorenz curve, but the or-dinate values are multiplied by the average of the x-values. This means that now the points with coordinates

, kskN N

,

k = 0, …, N are connected by straight line segments.

Just as the Lorenz curves introduce a partial order in the set of diversity arrays, also Shorrocks’ generalized Lorenz curves introduce a – finer – partial order. Accept-able diversity measures must then respect this partial order.

5. CONCLUSION AND SUGGESTIONS FOR FURTHER RESEARCH

In this note, I just want to put forward the idea that scale-invariance may not be a nec-essary requirement when using diversity measures to study interdisciplinarity via references. This leads to at least three ques-tions to be studied in further investigations:

1. Is the Shorrocks curve an acceptable way to study interdisciplinarity in a way that is not scale-invariant? Or should it be adapted?

2. How to include the other diversity requirements into this study? A first step could be to take disparity “seriously” as proposed in (Rousseau, 2018), while adapting the Leydesdorff-Wagner-Bornmann approach (Leydesdorff et al., 2019) or intrinsic diversity profiles (Patil & Taillie, 1979; Rousseau et al., 1999) are other alternatives.

I hope to be able to answer these questions in further investigations.


SHO

RT

COM

MU

NIC

ATI

ON

S, A

RTI

CLES

20

ACKNOWLEDGEMENT

I thank Lin Zhang (Wuhan University and KU Leuven) for helpful discussions.

REFERENCES

Hu, XJ., Rousseau, R., & Chen, J. (2011). On the defi-nition of forward and backword citation gen-erations. Journal of Informetrics, 5(1), 27-36.

Jost, J. (2006). Entropy and diversity. Oikos, 113(2), 363-375.

Jost, J. (2009). Mismeasuring biological diversity: Response to Hoffmann and Hoffmann (2008). Ecological Economics, 68(4), 925–928.

Leinster, T., & Cobbold, C.A. (2012). Measuring diversity: the importance of species similar-ity. Ecology, 93(3), 477– 489.

Leydesdorff, L., Wagner, C.S., & Bornmann, L. (2019). Diversity measurement: Steps to-wards the measurement of interdisciplinar-ity? Journal of Informetrics, 13(3), 904 -905.

Lorenz, M.O. (1905). Methods of measuring con-centration of wealth. Journal of the American Statistical Association, 9(70), 209-219.

Nijssen, D., Rousseau, R., & Van Hecke, P. (1998). The Lorenz curve: a graphical representa-tion of evenness. Coenoses, 13(1), 33-38.

Patil, G.P., & Taillie, C. (1979). An overview of diversity. In: J.F. Grassle, G.P. Patil, W. Smith and C. Taillie (Eds.), Ecological Diversity in Theory and Practice, International Co-operative Publishing House, Fairland, MD, pp. 3-27.

Rafols, I., & Meyer, M. (2010). Diversity and net-work coherence as indicators of interdisci-plinarity: Case studies in bionanoscience. Scientometrics, 82(2), 263–287.

Rousseau, R. (2018). A new method for diversity measurement: Taking similarity between

cells seriously. In: STI 2018 Conference Pro-ceedings, pp. 793-798.

Rousseau, R., Egghe, L., & Guns, R. (2018). Becom-ing metric-wise. A bibliometric guide for researchers. Oxford: Chandos (Elsevier).

Rousseau, R., Hu, XJ., & Zhang, L. (2019). Knowl-edge integration: its meaning and meas-urement. In: W. Glänzel, H.F. Moed, U. Schmoch, M. Thelwall (Eds.), Springer Handbook of Science and Technology Indica-tors (pp. 69-94). Cham: Springer.

Rousseau, R., & Van Hecke, P. (1999). Measuring biodiversity. Acta Biotheoretica, 47(1), 1-5.

Rousseau, R., Van Hecke, P., Nijssen, D., Bogaert, J. (1999). The relationship between diversity profiles, evenness and species richness based on partial ordering. Environmental and Eco-logical Statistics, 6(2), 211-223.

Shorrocks, A.F. (1983). Ranking income distribu-tions. Economica, 50(197), 3-17.

Stirling, A. (2007). A general framework for analysing di-versity in science, technology and society. Journal of the Royal Society Interface, 4(15), 707–719.

Wagner, C.S., Roessner, J.D., Bobb, K., Klein, J.T., Boyack, K.W., Keyton, J., Rafols, I., & Börner, K. (2011). Approaches to understanding and measuring interdisciplinary scientific research (IDR): A review of the literature. Journal of Informetrics, 5(1), 14 – 26.

Zhang, L., Rousseau, R., & Glänzel, W. (2016). Diversity of references as an indicator of the interdisciplinarity of journals: Taking simi-larity between subject fields into account. Journal of the Association for Information Sci-ence and Technology 67(5), 1257–1265.

Zitt, M., Lelu, A., Cadot, M., & Cabanac, G. (2019). Bib-liometric delineation of scientific fields. In: W. Glänzel, H.F. Moed, U. Schmoch, M. Thelwall (Eds.), Springer Handbook of Science and Tech-nology Indicators (pp. 25-68). Cham: Springer.


SHO

RT

COM

MU

NIC

ATI

ON

S, A

RTI

CLES

21

A MULTILAYER NETWORK ANALYSIS OF MOBILITY AND COLLABORATION IN EUROPEAN UNIVERSITY RESEARCH

1 ECOOM, KU Leuven, Leuven, Belgium2 MTA-PE Budapest Ranking Research Group, University of Pannonia, Budapest, Hungary3 Wigner Research Center for Physics,

VIAS – Virtual Institute of Advanced Studies, Budapest, Hungary4 Library and Information Centre of the Hungarian Academy of Sciences,

Dept Science Policy & Scientometrics, Budapest, Hungary

BART THIJS 1

ANDRÁS TELCS 2, 3

WOLFGANG GLÄNZEL 1, 4

PEI-SHAN CHI 1

ZSOLT KOSZTYÁN 2

LÁSZLÓ GADÁR 2

VIVIAN CSÁNYI 2

ABSTRACT: The present report gives an overview of a recent joint Hungarian-Flemish project funded by the Research Foundation - Flanders (FWO). The project aimed to study European university research from the viewpoint of scientists’ mobility and collaboration patterns of the most important universities of selected countries.


SHO

RT

COM

MU

NIC

ATI

ON

S, A

RTI

CLES

22

INTRODUCTION

A two-year joint research project between the ECOOM team at KU Leuven and the Budapest Ranking Research Group at Hungarian Academy of Sciences were con-ducted during 2018 and 2019. This project aims to provide a high level of analytic and visualization system to explore the scien-tific contribution of European universities. Based on its capacities the parties studied and developed fine-tuned, targeted uni-versity multilayer networks. In short, the partners created a database which is basi-cally a multi-labelled graph in which verti-ces are institutions and edges are citations. Institutions carry some feature descripting labels while edges carry labels describing the citation relation in details (publication time, citation time, subject classification etc.). Unlike other endeavours, this pro-ject did not aim to conduct any ranking exercise, which provided us the liberty to choose appropriate data sources and to se-lect European countries and universities for a more complex multilayer exercise. In par-ticular, we have selected the following data sources to highlight on essential aspects in researchers’ mobility and collaboration.

The Erasmus mobility network – used to investigate spatial characteristics for geo-coded institutes – was enriched with the integration of mobility data and three data-bases: the institutional socio-economic data of the European Tertiary Education Regis-ter (ETER)1, the Global Research Identifier Database (GRID)2 as well as points of inter-est concerning surrounding institutions. The network provides relevant information of the mobility patterns and the attractive-ness of the institutions. A publication net-work for the 2013 data with weighted links was based on bibliographic coupling and conduced to further university clustering. The Academic Space Explorer (ASE) system was then adapted with some filtering, clus-

1 https://www.eter-project.com/2 https://www.grid.ac/

tering and analytic tools to illustrate the evaluative profiles of European countries and the interconnections between univer-sities in Belgium, Germany and Hungary.

RESULT SUMMARY

The project cooperated by two institu-tions, KU Leuven and Hungarian Academy of Sciences, generated several national and international research outputs, for a na-tional conference (Csányi, 2019), three con-tributions for international conferences (Kosztyán et al., 2018a, 2018b; Fehérvölgyi, et al, 2019), four papers for international peer-reviewed journals (Gadár et al., 2018; Gadár & Abonyi, 2019; Gadár et al., 2019; Kosztyán et al., 2019) and six research semi-nars organised in Budapest and Leuven. In what follows, we introduce and summarise the main findings of these results.

The Erasmus activity network served as the main source of tracking mobility and analysing European networks from this perspective. The student and teacher mo-bility patterns based on 2013/2014 Erasmus data are somewhat different (see Figure 1; top: students, bottom: teachers). This is to a certain extent due to the different mo-tivations for and durations of their stays abroad. Generally Spanish universities at-tract more students and academic teachers compared to their outgoing ones. By con-trast, Turkish and Polish universities have a disproportionate number of outgoing teachers as compared with the incoming ones, whereas German and Italian univer-sities have much more outgoing students than incoming ones. Figure 1 also illus-trates the active participation of Spanish universities in the Erasmus project.

In order to gain a deeper insight into the Erasmus student mobility, activities in three different subject areas were stud-ied (cf. Figure 2). We have selected applied fields in the sciences along with the hu-manities based on Erasmus project clas-sification. In particular, in Social science,

https://www.eter-project.com/

https://www.grid.ac/

https://link.springer.com/article/10.1007/s11233-019-09028-x#auth-3


SHO

RT

COM

MU

NIC

ATI

ON

S, A

RTI

CLES

23

0 500 1000 1500

Num

ber o

f inc

omin

g st

uden

ts in

201

3/20

14 1500

1000

500

0

Number of outgoing students in 2013/2014

ALL SUBJECT

0 100 200

Num

ber o

f inc

omin

g ac

adem

ic te

ache

rs in

201

3/20

14

200

100

0

Number of outgoing academic teachers in 2013/2014

ALL SUBJECT

Figure 1. Scatter plots of 2013/2014 Erasmus incoming vs. outgoing staff. Top: students; bottom: academic teachers


SHO

RT

COM

MU

NIC

ATI

ON

S, A

RTI

CLES

24

Business and Law, Spanish uni-versities attract more students from other universities than their own students who leave for staying at other institutions abroad. In Science, Mathemat-ics an Computing, Nordic and some Belgian and Dutch uni-versities show their advantage to attract more students from other universities than send-ing out their own students. In Health and Welfare, a few uni-versities from Italy, Czech Re-public, Sweden, Denmark and Hungary show their receiving role for unproportionally more incoming than outgoing stu-dents. The observed subject-related difference revealed in Figure 2 points us a new direc-tion to dig in from the mobility data. We further looked at the spatial multiplex network of the student mobility based on the determined GPS coordinates in the three main subject areas.

Figure 3 – inspired by Ga dár et al. (2019) – linked the data from the Erasmus project, the European Tertiary Education Register (ETER) and the Global Research Identifier Database (GRID) to convert a spatial net-work on geographical map. It shows different connections in the three subject areas. In So-cial science, Business and Law, the connections between the UK and European countries are strikingly strong. Universities in Spain, France, Germany, Bel-gium and Italy are also impor-tant contributors in the Erasmus student mobility network. In the field Science, Mathematics and Computing, Nordic universities play a more important role in the network. Finally, in Health and

Figure 2. Scatter plots of 2013/2014 Erasmus incoming VS. outgoing stu-dents in different fields. Top: Social science, Business and Law; Middle: Science, Mathematics and Computing; Bottom: Health and Welfare

800

600

400

200

00 200 400 600 800


Num

ber o

f inc

omin

g st

uden

ts in

201

3/20

14

SOCIAL SCIENCES, BUSINESS AND LAW

200

150

100

50

00 50 100 150 200 250


Num

ber o

f inc

omin

g st

uden

ts in

201

3/20

14

HEALTH AND WELFARE

250

200

150

100

50

00 50 100 150 200 250


Num

ber o

f inc

omin

g st

uden

ts in

201

3/20

14

SCIENCE, MATHEMATICS AND COMPUTING


SHO

RT

COM

MU

NIC

ATI

ON

S, A

RTI

CLES

25

Welfare sciences, The Netherlands is send-ing a large number of students to Denmark. Italy, Spain and Czech Republic are central nodes of student mobility in this field.

A further perspective to focus on the countries we are especially interested in, was taken and presented in Figure 4. This new dimension was added on the basis of scholarly collaboration measured by co-authorship of scientific publications in-dexed in Clarivate Analytics Web of Sci-ence Core Collection (WoS). Although the time frames and actors are different in these dimensions, i.e., we have not studied the publication patterns of the Erasmus students but the complete research staff affiliated with the selected universities, at this level of aggregation this approach may add a new dimension not directly linked to the network created by previous actors. To zoom into targeted countries, the student and teacher mobility and scholarly publica-tion collaboration networks were limited into the international connections among Belgium, Germany and Hungary. Regard-ing the student mobility, universities in Berlin are highly connected with Flemish universities and universities in Budapest. Universities in Cologne area and Munich area received and sent out many students from or to universities in Budapest and Flanders. The teacher mobility was more equally distributed among universities. The connections from Hungary are not as strong as those among Belgium and Ger-many. In terms of WoS publications in 2013, Belgian universities strongly collabo-rated with the other two countries, espe-cially with German universities.

DISCUSSION AND FUTURE RESEARCH

The joint project has offered a new per-spective. The usual macro and meso bib-liometric study of research collaboration and the socio-bibliometric analysis of stu-dents’ and researchers’ mobility could be

Figure 3. Mobility networks with different subject dimensions in the 2013-2014 academic year similarly to Gadár et al. (2019). The size and colour intensity of the nodes are proportional to the number of in-coming students. The size and colour intensity of the edges are proportional to the number of students that travel between the connected HEIs. Top: Social science, Business and Law; Middle: Science, Math-ematics and Computing; Bottom: Health and Welfare


SHO

RT

COM

MU

NIC

ATI

ON

S, A

RTI

CLES

26

combined into a first mul-ti-layer framework, even if only a small number of institutions could be taken into account. Overlapping network layers indicate how correlated (or not) mobility and research col-laboration systems are with each other. Based on our preliminary results, there does not appear to be a significant overlap between mobility and re-search collaborations, due to the different motiva-tions and driving forces of different collaboration.

In the present project we could much benefit from our local knowledge and experience, which was all the more important because we had to clean a large amount of data on the bibliometric part for the integration into the framework. The main ob-jective and achievement of the project was there-fore rather the creation of the framework and the elaboration of the possible components of this multi-layer network. Figure 4 has shown that different patterns can be attributed to individual components. Beyond the programmes enabling student and staff mobility, there are differ-ent individual interests in-fluencing and shaping the patterns of student mo-bility particular making destinations attractive on the long-term (education, career) compared to rather short-term teacher mobil-

Figure 4. Mobility and scholarly collaboration international networks be-tween Belgium, Germany and Hungary. Top: students; MIddle: academic teachers; Bottom: WoS publications


SHO

RT

COM

MU

NIC

ATI

ON

S, A

RTI

CLES

27

ity. Collaboration patterns are, by contrast, influenced to a lesser extent by individual interests as those are part of larger, mostly subject-specific, European and even global collaboration frameworks. These patterns can be recognised in Figure 4. The sparser mobility networks top and notably in the centre is contrasted by the denser col-laboration networks that are often based multilateral projects. This also implies the involvement of more universities and the higher density of networks. At the future works, subnetworks of subjects can be compared. In addition, further layers can be integrated to the multilayer structure such as co-authorship network of publica-tions and the patents. In this way the mo-bility and collaboration activities of HEIs can be deeper analysed.

A further recognition in the project is that we found different network patterns in subject area layers, which is mostly re-lated to the characteristics of the institu-tions. In case of real sciences, the attrac-tiveness of Northern European universities for student mobility is apparent, while in other subject areas Southern European in-stitutions receive the most students. The characteristics of institutions depend on the representation of disciplines moreo-ver the educational and/or research and/or innovation focus of the institution. Dif-ferent characteristics result different col-laboration systems resulting different net-work patterns. Future research directions could be the examination of correlations between the strength of institutions, the position of higher education rankings and the characteristics of the emerged collabo-ration networks.

ACKNOWLEDGMENTS

The project was partially funded by “Analysis of Scientific Performance of Universities" – bi-lateral mobility project of HAS – FWO 2018-19. A bilateral collaboration programme of the Research Foundation – Flanders (FWO).

REFERENCES

Fehérvölgyi, B., Kosztyán, Z., Pál, M., Csizmadia, T., Kerekes, K. (2019). Measuring third mission activities of the universities by multi-layer networks. The 4th International Conference on Economics and Business Management – ICEBM, 23 November, Cluj-Napoca, Romania.

Csányi, V. (2019). The position of Hungary in in-ternational patent statistics. LXI. Georgikon Conference. 3-4 October, Keszthely, Hungary.

Gadár, L., Abony, J. (2019). Frequent pattern min-ing in multidimensional organizational networks. Scientific Reports, 9, Article number 3322.

Gadár, L., Kosztyán, Z., Abony, J. (2018). The Set-tlement Structure Is Reflected in Personal Investments: Distance-Dependent Network Modularity-Based Measurement of Regional Attractiveness. Complexity, Article ID 1306704, 16 pages.

Gadár, L., Kosztyán, Z., Telcs, A., Abony, J. (2019). A multilayer and spatial description of the Eras-mus mobility network. Scientific Data, 7, 41.

Kosztyán, Z., Banász, Z., Csányi, V., Telcs, A. (2019). Rankings or leagues or rankings on leagues? – Ranking in fair reference groups. Tertiary Edu-cation and Management volume, 25, 289–310.

Kosztyán, Z., Gadár, L., Telcs, A. (2018). Student and Teaching Mobility & Knowledge Trans-fer: a social network analysis study. The Third University Mission Conference, 30 November, Moscow, Russia.

Kosztyán, Z., Banász, Z., Csányi, V., Telcs, A. (2018). From Global Rankings to Rankings on Leagues – a case study of country leagues. IREG-9 Conference, 23-25 May, Hasselt, Belgium.




editorial - issi-society.orgissi-society.org/media/1407/newsletter61.pdf · editorial jointly study...

Documents