CC BY
39
Irina Marshakova-Shaikevich
The Institute of Philosophy of RAS Moscow, Russia AM University Poznan, Poland e-mail: ishaikev@mail.ru
Bibliometrik Maps of Scientific Collaboration of EU in Science and Social Science1
The paper presents bibliometric analysis of the international scientific collaboration of 27 EU countries in the field of science and social sciences. The object of the paper is the joint publications of authors from the 27 EU member-states in the fields of science and of social sciences. Material for this analysis was drawn from DBs WOS: SCIex 2006 and SSCI 2006. The states as wholes are considered here as participants of collaboration. If a paper has two or more authors from different countries, it belongs to the material of the present study. The discovery of specific country-to-country links is based on the comparison of actual collaboration figures with theoretical values, calculated on the assumption of mutual statistical independence. The results are presented with the help of two maps showing those specific links.
Key words: map of science, bibliometric analysis, international collaboration, EU countries, Web of Science, SSCI CDE.
Dedicated to Vladislav Kelle
Introduction
Scientific collaboration is becoming a significant issue in bibliometric and webomet-ric studies of science. Only some of them should be mentioned in the context of this study: Lamirel et al. presented a new approach for evaluation of collaboration between European universities (Lamirel, 2005). Science collaboration of in the field of social sciences was discussed by the present author at Collnet 2006 in Nancy (Marshakova-Shaikevich 2006, 2007), bibliometric perspectives of an integrated EU research was presented by Robert J.W. Tijssen at Collnet 2008 in Berlin (Tijssen, 2008). An attempt at visualization of international collaboration of 27 EU countries in science and social scinces is made here. The resulting maps could serve for better understanding of structure of scientific cooperation in Europe.
The main object of the present study is international scientific collaboration within the EU countries. However, lest essential information should be lost, the data for some countries outside European Union were also included into material of this study. Those countries are — USA, Japan, People’s Republic of China, Canada, Australia, India, Russia, Switzerland, Israel, Norway, Iran, Croatia and Iceland. The statistics of academic publications of 40 states under study are given in Table 1.
1 Доклад, представленный на 5th International Conference on Webometrics, Informetrics and Scientometrics & 10th COLLNET Meeting, 13—16 September 2009, Dalian, China.
Countries
Table 1. Research activity in science and social sciences: 2006
Social Sciences: Total N = 150th Science: World Total N= 1218 th
Country SSCI 2006 Number publications Weight in world total, % SCIex 2006 Number of publications Weight in world total, %
Austria 722 0.48 10750 0.88
Belgium 1465 0.98 15471 1.27
Denmark 998 0.66 11047 0.91
Finland 1111 0.74 9556 0.78
France 2934 1.96 61899 5.08
Germany 6928 4.62 89306 7.33
Greece 593 0.39 10463 0.86
Ireland 654 0.44 7067 0.56
Italy 2441 1.63 50886 4.18
Luxembourg 0 265 0.02
Netherlands 4725 3.15 28315 2.32
Portugal 405 0.27 7488 0.61
Spain 2855 1.90 37808 3.10
Sweden 2290 1.53 19466 1.60
UK 19691* 13.13 100575* 8.52
CzR 270 0.18 7203 0.59
PL 403 0.27 17021 1.40
Hu 317 0.21 5974 0.49
SK 151 0.10 2305 0.19
Slovenia 124 0.08 2225 0.18
Estonia 88 0.06 874 0.07
CYPRUS 57 0.04 364 0.03
Lithuania 107 0.07 1213 0.10
MALTA 14 0.01 72 0.01
LA 20 0.01 343 0.03
Romania 70 0.05 3219 0.26
Bulgaria 51 0.03 1863 0.15
USA 74567 49.7 376833 30.94
Australia 6676 4.45 32718 2.69
Canada 8951 5.97 51725 4.25
Russia 530 0.35 23188 1.90
Croatia 334 0.22 2118 0.17
Israel 1669 1.11 14310 1.17
Japan 1828 1.21 88851 7.29
Norway 1148 0.76 7712 0.63
Switzerland 1808 1.20 21055 1.73
Iceland 82 0.05 574 0.05
Peoples-R-China 1965 1.3 86025 7.06
*The sum of UK publications adjusted for “internal collaboration”
The EU countries contribute some 40 % to the world output in science proper and 33 % in social sciences.
Bibliographic data were drawn from DBs WOS SCIex 2006 and SSCI 2006.
The source data call for some modifications for our purpose. The tradition of British authors to indicate parts of UK (England, Scotland, Wales, Northern Ireland) rather than the kingdom as a whole gives those parts undue prominence. The figures of that “internal collaboration” should be subtracted from the sum of four parts to give the true result for UK.
Methodology
To calculate specific collaboration relatedness of two countries the following formula was used :
Sij = (C j' — m) / V m
Cij — real number ofjoint publications of countries i and j
m — mathematical expectation of number ofjoint publications of countries i and j
The mathematical expectation of the number of joint publications, entirely due to chance, was calculated as
m = Ni x Pj,
with Ni — total collaboration publications of country i ,
Pj = Nj / £ N — the weight of country j in the total number of publications of the world.
An example of seven European countries is given in Table 2 and Table 3 to illustrate the calculation. In Table 2 the figures below the main diagonal indicate mathematical expectations (m) and those above the main diagonal are real numbers ofjoint publications (C).
Table 2
Countries France Belgium Netherlands Germany Switzerland Austria Italy
France XXXX 1876 1742 4536 2251 588 3383
Belgium 784 XXXX 1478 1433 608 354 940
Netherlands 1437 358 XXXX 3168 912 370 1402
Germany 4535 1133 2067 XXXX 3572 2200 3434
Switzerland 1069 268 490 1547 XXXX 599 1523
Austria 551 135 247 790 186 XXXX 634
Italy 2583 646 1183 3732 882 449 XXXX
The resulting measure of specific collaboration relatedness (S) is given below the main diagonal of Table 3.
Table 3
Countries France Belgium Netherlands Germany Switzerland Austria Italy
France XXXX 1876 1742 4536 2251 588 3383
Belgium 39 XXXX 1478 1433 608 354 940
Netherlands 8 58 XXXX 3168 912 370 1402
Germany 0 8 22 XXXX 3572 2200 3434
Switzerland 36 20 19 51 XXXX 599 1523
Austria 1 10 7 50 30 XXXX 634
Italy 15 8 6 -4 21 8 XXXX
One can see that the of French and German joint publications corresponds exactly to what might be expected as a null-hypothesis, i.e. those countries are statistically independent as far as international collaboration is concerned. S = 0 here. The number of German and Italian joint publications is even less than expected (S = -4), which might be interpreted as the effect of a very weak factor hindering the collaboration between the two countries. On the other hand, all other cells are positive deviations from mathematical expectation. In the pair “France-Aus-tria” this association is negligible, in all other pairs it is significant or highly significant.
A matrix of 27 x 27 would be too cumbersome. A graphic representation is easily visualized and gives better ground for interpretation.
Discussion and conclusion
In Fig. 1 a map of international collaboration in science is given. The four degrees of thickness of the lines reflect values of S as follows:
S - 50 and more (e.g. UK — Ir, CzR — Sk, Au — Ge),
S = 25-49 (e.g. Fr — Swt, De — Ic, Fi — Es, Sl - Cr),
S = 15-24 (e.g. Fr — It, Ge — Ru, CzR — Pl, Hu — Ro),
S = 10-11 (e.g. UK — M, Sp — It, Bu — Gr, Sw — Li).
Fig. 1. Map of collaboration in science
A very compact cluster of countries is seen in the upper part of the map: it comprises five Scandinavian countries with three small Baltic states. One bridge to the rest of Europe is formed by United Kingdom with three states: Australia, Ireland and Malta. Another bridge (of weaker S) connects Scandinavia with Germany, Netherlands and Belgium. Specific links of the countries of Romance languages are manifest in the left part of the map. Switzerland with its languages and its scientific tradition is a virtual hub of the map. It has strongest ties with Germany and Austria, who show various specific links with the states of Central and Eastern Europe (Russia include). Some of those countries (especially Hungary and Bulgaria) are apt to participate in international collaboration to an uncommon degree. In addition to the links of the map we can mention Hungary’s links with Finland, Sweden and Poland, or Bulgaria’s ties with France, Italy, Austria and Romania. On the contrary, Poland’s involvement this cluster is not so great, although its weak ties (S = 6-8) with Austria, Germany, Russia and Lithuania might be mentioned.
No specific links were observed outside Europe (with the single exception of ‘ UK-Aus-tralia’ pair).
In Fig. 2 a map of the international collaboration in social science is given Specific links in social sciences to some extent resemble the outlines of Fig. 1 on a lesser scale. The output
Fig. 2 Map of collaboration in social science
of publications in social sciences is eight times less than that in science proper. Accordingly the values of S are much lower. The four degrees of thickness of the lines reflect values of S as follows:
S — 20 and more (e.g. Be — Ne, CzR — Sk, Es — La),
S = 15-19 (e.g. Ge — Swt, Li — Es, Sl - Cr),
S = 9-14 (e.g. Sw — No, Ge — Au, CzR — Pl, Hu — Bu),
S = 4-8 (e.g. De — Fi, Sp — It, Fr — Swt Sw — Li).
United Kingdom disappears from the map altogether. Scando-Baltic cluster remains intact. The linguistic preponderances are evident in Western Europe (with German and Romance languages very active as cultural factor). But East European cluster with its Soviet and more ancient Austro-Hungarian traditions is alive.
In conclusion I’d like to add that there is no connection between USA and EU countries, USA only have a weak links with Norway (S=6) and Israel (S=3). China appears quite isolated. It does not show any links to the countries under study.
References:
J.C. Lamirel, S.Al. Shehabi, C. Francois, 2005: Evaluation of collaboration between European universities using dynamic interaction between multiple sources. ISSI 2005: Proceedings of the 10th International Conference of the International Society for Scientometrics and Informetrics, Vols 1 and 2. 2005. p. 740-749 Karolinska Univ Press AB, Stockholm.
Irina Marshakova-Shaikevich, 2007: The visualization of scientific collaboration of 15 “old” and 10 “new” EU countries in the field of social sciences // Collnet Journal of Scientometrics and Information Management. Taru Publication, New Delhi (ISSN: 0973-7766), Vol.1, N1, June 2007. P. 9-16.
Irina Marshakova-Shaikevich, 2006: Science collaboration of new 10 EU countries in the field of social sciences. Information Processing & Management 42 (2006) 1592-1598.
Robert J.W. Tijssen, 2008: Are we moving towards an integrated European Research Area? Some macro-level bibliometric perspectives. H. Kretschmer & F. Havemann (Eds.): Proceedings of WIS 2008, Berlin Fourth International Conference on Webometrics, Informetrics and Scientometrics & Ninth COLLNET Meeting Humboldt-Universitat zu Berlin. http://www.collnet.de/Berlin-2008/ Proceedings-WIS-2008.pdf.
Web of Science: 2006.
