Mapping international collaboration in science in Asia through coauthorship analysis

(Dedicated to Dr Eugene Garfield, scientometricist par excellence, on his 75th birthday)

Subbiah Arunachalam
M S Swaminathan Research Foundation, Third Cross Street, Taramani Institutional Area, Chennai 600 113, India, email: arun@mssrf.res.in

M Jinandra Doss
Central Leather Research Institute, CSIR, Adyar, Chennai 600 020, India


Abstract


Using coauthorship data from the CD-ROM edition of Science Citation Index 1998, we have analysed international collaboration in science in 11 Asian countries, viz. India, China, Japan, Taiwan, Hong Kong, South Korea, Singapore, Thailand, Malaysia, Philippines, and Indonesia. We have looked at papers resulting from collaboration between these countries and G7, European Union and OECD countries as well as among themselves and with selected Latin American and African countries. We have computed internationalization index for all these countries as well as international cooperation index and affinity index for cooperation between these countries and countries collaborating most often with them. We have classified all the papers from all these countries including those resulting from international collaboration with many countries into about 25 standard subject categories (provided by ISI) such as physics, chemistry, engineering, and biology & biochemistry as well as into different journal impact factor ranges. These data have also been aggregated into seven major subject categories. We have used the sum of impact factors of all papers as well as papers under major subject categories to characterize each country's total and collaborated scientific literature output. Among the three Asian giants, Japan (16.4% of papers internationally collaborated), India (17.6%) and Taiwan (16.3%) recorded an internationalization index less than 30 whereas China (28.5% of papers), South Korea (24.6%), and Hong Kong (36.2%) recorded an internationalization index greater than 40. The smaller Asean countries have recorded much higher internationalization indices - 130 for Indonesia (88% of papers), 112 for Philippines (67% of papers) and 83.9 for Thailand (59.4% of papers). India, China and South Korea have collaborated more in physics, whereas the other eight countries had collaborated more in life sciences. These are the two areas where multinational collaboration is higher than in other areas in most Asian countries. In general, internationally coauthored papers appear in higher impact factor journals. In almost all fields and for all 11 Asian countries, USA is the most preferred collaborating partner. All G7 countries collaborate more with China, which is emerging as a leader in regional collaboration, than with India. Within Asia, the two largest collaborations are between China and Japan (627 papers) and between Japan and South Korea (416 papers). India's largest Asian partner is also Japan (195 papers).


Introduction


Science is becoming increasingly global as reflected by the rapid growth in international collaboration.(Ref. 1-3) The past decade witnessed a vigorous increase in international collaboration, as indicated by internationally coauthored scientific and technical articles. From 1986-88 to 1995-97, the total number of articles in the ISI databases increased by 12%; multi-institutionally coauthored papers by 46% (from an annual average of 177,100 to 258,500); and internationally coauthored articles increased by almost 115% (from 35,700 to 76,200).(Ref. 4) In 1997, half of the papers in a set of about 5,000 journals indexed in Science Citation Index since 1985 had authors from more than one institution and 30% of these coauthored articles involved international collaboration, compared to 23% a decade earlier.(Ref. 4) The growing international nature of science is further underscored by the increase in the share of cross-national citations; in 1997, about 59% of all citations were to non-domestic papers, as against 53% in the early 1990s.(Ref. 4) Traditionally, most of these international collaborations have been taking place in the advanced (G7, EU and OECD) countries, many of which also publish large numbers of research papers.(Ref. 5,6) But in recent years, Asian countries are fast increasing their share of worldwide international collaboration in science. It is this that we have attempted to map in this paper.
This paper is an extension of an earlier paper by Arunachalam where he has used international coauthorship data taken from SCI 1998 to map international collaboration involving India, China, Japan and Israel.(Ref. 7) A few years ago Arunachalam et al.(Ref. 8) had mapped international collaboration involving India, China and Japan using data from SCI 1991. Using similar methodology, Arunachalam and Jinandra Doss(Ref. 9) have mapped international collaboration in new biology research in Israel using data from three years of Biochemistry and Biophysics Citation Index (1992, 1995 and 1998).


Methodology


We downloaded, in comma delimited form, all papers indexed in the CD-ROM version of SCI 1998 with an address in anyone of the following countries: India (IN), China (CN), Japan (JP), South Korea (KR), Taiwan (TW), Hong Kong (HK), Singapore (SG), Thailand (TH), Malaysia (MY), Indonesia (ID), and Philippines (PH). Only (full-length research) articles, reviews and notes, were taken into account, as these tend to report substantive research results. We used a custom format covering the following fields: Name(s) of author(s), author address(es), source (journal title, volume, year, and page), document type, ISSN, and language. For each journal title, we added the impact factor value from Journal Citation Reports 1997 (CD-ROM edition), and the standard subject category from information obtained from the Research Group of the Institute for Scientific Information. ISI has categorized all journals into about 25 standard subject categories and more than 100 deluxe categories. We have classified journals (and papers published in them) into the standard subject categories and also grouped them into seven larger groups. We used FoxPro and Visual Basic for analyzing the data.

We identified papers having more than one country in the address field and calculated for each country its internationalization index, based on the number of international linkages, as suggested by Davidson Frame and Carpenter(Ref. 10):

Internationalization Index = 100 x number of international links/ total number of papers from the country

We calculated the (international) cooperation index and affinity index from papers resulting from collaboration between each one of the 11 countries with many selected countries, following the procedure described earlier.(Ref. 7) We have also classified internationally coauthored papers into major subject categories, based on journal titles, and calculated average expected impact factors for each subject category as well as for the entire collaborated output.


Results and Discussion


In Table 1 we give the total number of papers from these 11 countries, the number of papers resulting from international collaboration, the total number of international linkages and the number of papers involving different number of collaborating countries. Japan, with over 60,000 papers, is second only to the USA in the number of papers published. China (11th), India (14th) and South Korea (15th) are the only three other Asian countries to figure in the top 15 producers of scientific papers. Whereas about 17% of papers from India, Japan and Taiwan have resulted from international collaboration, close to one-fourth of Korean papers, over 28% of Chinese papers and over 30% of papers of Singapore have resulted from international collaboration. Countries with very small publication output, such as Philippines (67%) and Indonesia (88%), have recorded a very high percentage of internationalization.

For all countries, international collaboration involving one other country accounts for bulk of the collaboration. That is followed by collaboration involving three countries. Multinational collaboration involving more than three countries is relatively rare, except in the case of Japan. Following Miquel,(Ref. 11) we have made a distinction between the number of collaborative papers and total number of collaborative links. More the number of countries in the byline of papers from a country, the greater will be the number of links.(Ref. 7)

Table 2 lists the number of papers in which the eleven Asian countries have collaborated with about 50 countries, including the G7 countries, the European Union countries, OECD countries, the newly emerging economies of Asia, the Tigers, Israel and selected African and Latin American countries. Please note that in some of these papers there may be collaborators from more than the two countries listed. Except for Malaysia, for every country considered, the United States is the preferred partner for collaborative research. Worldwide, 44% of all internationally coauthored papers published in 1995-97 had at least one US coauthor.(Ref. 4) With a few exceptions, from 25 to 33% of European countries' internationally coauthored papers involved collaboration with the United States. For major science-producing Asian nations, coauthorship with US researchers ranked higher. Japan and India - both nations with relatively low overall rates of international collaboration - shared 46 and 40%, respectively of their internationally coauthored articles with United States researchers. For other Asian nations, collaboration rates with the United States ranged from 70% for Taiwan to 31% for Singapore (Ref.4, p. 6-49).

In general, the 11 Asian countries collaborate often with G7 countries and to some extent with Australia. China has collaborated with Australia in 171 papers, with the Netherlands in 107 papers, with Spain in 98 papers and with Sweden in 81 papers. India has collaborated with Australia in 80 papers, with Switzerland in 76 papers, with Spain in 73 papers and with the Netherlands in 59 papers. Indonesia has collaborated with Australia in 40 papers and with the Netherlands in 26 papers, and Singapore has collaborated with Australia in 63 papers. South Korea has collaborated with Spain in 80 papers, with Switzerland in 69 papers and with Poland in 46 papers. The Asian countries hardly collaborate with countries in Latin America and Africa.

Table 3 lists the cooperation index and affinity index for collaboration between the Asian countries on the one hand and their major collaborating partner countries. China has a much higher cooperation index with G7 countries, South Korea, Taiwan, Hong Kong and Singapore than India. Japan has a higher cooperation index with other G7 countries than China. Malaysia, which has published only 538 papers, has a cooperation index of over 20 with both Singapore and Thailand, over 14 with the UK and over 10 with India. With the exception of Malaysia, all the countries have a high affinity towards the USA. Malaysia has a higher affinity to the UK.

Although, compared to collaboration among the countries of the European Union (Ref. 7), the level of intra-Asian research collaboration is rather low, it has increased considerably in recent years (Table 4). The main trend in Asia appears to be the development of regional collaborative patterns involving China and the newly industrialized economies. Overall, intraregional collaboration increased from 15% of all Asian foreign collaboration in the late 1980s to 24% a decade later.(Ref. 4) For China, intraregional collaboration rose from 15.8 to 35% of its internationally coauthored papers; and for Singapore from 19.5% to 37%. However, regional collaboration remained relatively low for Japan and India, at about 15% of their internationally coauthored articles in 1995-97 (Ref. 4, p. 6-50).

An interesting observation is that virtually every country considered here collaborates with a larger number of countries now than a few years ago. In 1986-88, India had collaborated with 87 countries, and the number rose to 109 in 1995-97.(Ref. 4) Corresponding figures for other Asian countries are(Ref. 4):

Country No. of collaborating countries
  1986-88 1995-97
China 54 107
Japan 80 127
South Korea 33 83
Taiwan 34 67
Hong Kong 35 82
Singapore 42 65
Thailand 49 94
Malaysia 32 76
Indonesia 39 84
Philippines 44 75

In Table 5 we provide data on the percent of papers resulting from international collaboration and the average expected impact factor of internationally coauthored papers from the 11 countries in journals in seven major fields, viz. physics, chemistry, life sciences, mathematics, engineering, geosciences and multidisciplinary journals. For every country, the average expected impact factor of collaborated papers is higher than the average expected impact factor of all papers as well as the average impact factor of non-collaborated papers (Figure 1). Whereas India, China, and South Korea have collaborated more often in physics, the other eight countries have predominantly collaborated in life sciences. Indeed, life sciences account for more than half of the collaborated papers of Thailand, Malaysia, Indonesia and Philippines (Table 5). Incidentally, the United States collaborates largely in the area of clinical medicine, a part of life sciences. India and Korea have roughly the same number of internationally coauthored papers, but there is considerable difference in the distribution of the collaborated papers by subject. Whereas India has collaborated with other countries more often than South Korea in physics, chemistry and geosciences, South Korea has collaborated more often than India in life sciences, mathematics and engineering.

In Table 6 we provide some statistics on the extent and distribution of internationally coauthored papers of South Korea, as well as data on the expected impact of these papers based on impact factors of journals in which the papers were published. Also given in Table 6 are the cooperation index and affinity index for papers resulting from South Korea's collaboration with eight major partner countries. USA dominates international collaboration with South Korea; indeed the number of papers resulting from US-Korea collaboration exceeds the number resulting from Korea's collaboration with the next seven leading collaborating countries! The affinity index for South Korea ? USA is close to 35 and the cooperation index is close to 30. Apart from the G7 countries, China is an important collaborator of South Korea. Invariably, the average expected impact factor of internationally coauthored papers is considerably higher than the average expected impact factor of all South Korean papers. Whereas over 48% of all papers from South Korea have appeared in journals of impact factor less than unity, among internationally coauthored papers only a smaller fraction has appeared in such low-impact journals. Again, of all papers from South Korea less than 12% have appeared in journals of impact factor greater than 3.0, but a much higher share of internationally collaborated papers have appeared in such higher-impact journals. In particular, more than 55% of papers resulting from collaboration between Korean authors on the one hand and Italian or French authors on the other have appeared in journals of impact factor greater than 3.0.

Table 7 gives similar data for internationally collaborated papers of Taiwan. For Taiwan also, China is a leading collaborating nation apart from the G7 nations. International collaboration in Taiwan is also dominated by the USA, which has collaborated with Taiwan in more papers than the next seven leading collaborating countries put together. The affinity index for Taiwan ? USA is close to 38 and the cooperation index is over 19. As with South Korea, here again, the expected average impact factor of internationally coauthored papers is much higher than that of all papers from Taiwan. In particular, the expected average impact of papers written in collaboration with the People's Republic of China is twice that of all papers from Taiwan. Whereas over 45% of all papers from Taiwan have appeared in journals of impact factor less than unity, a much smaller proportion of internationally collaborated papers - as low as 14% in the case of French-Taiwanese collaboration - has appeared in such low-impact journals. Less than 11.5% of all Taiwanese papers have appeared in journals of impact factor greater than 3.0, but a much higher percentage of internationally collaborated papers have appeared in such higher-impact journals. The figures vary from 22.2% for papers with Japan and 24.8% for papers with USA to 83.3% for papers with Italy and 56.9% for papers with France.

For India and China also USA is the number one collaborator, but USA is not as dominating as in the case of South Korea and Taiwan. Also, a very high percent of papers from these two countries, over 55% for India and over 53% for China, have appeared in journals of impact factor less than unity. A very low percent of papers have appeared in journals of impact factor higher than 3.0 - about 7.5% for India and about 6.8% for China. Here again, a much higher percent of internationally collaborated papers have appeared in high-impact journals.7 In the case of Japan the distribution of international collaboration is somewhat less even, with the US share among the top eight collaborators being 49.1% as against less than 40% for India and less than 35% for China.

Table 8 gives data on the distribution of internationally collaborated papers of South Korea by subject and collaborating country. South Korea collaborates often in physics (27.4% papers are internationally collaborated) and chemistry (16.5%) and to a lesser extent in engineering (20.1%) and clinical medicine (23.8%). Among countries its preferred partners are USA (more than 60% of collaborations are with the USA) and Japan (20.6%). The numbers in the top two rows - physics and chemistry - under all countries and most numbers under USA and Japan are larger than the rest of the numbers in the matrix.
Singapore collaborates most often with the USA (28% of all collaborated papers) and China (18.2%). About 37% of papers in physics, 32% of papers in chemistry, over 23.3% of papers in materials science, 23.5% of papers in engineering, and 22.4% of papers in clinical medicine from Singapore have come out of international collaboration (Table 9). Most of the numbers in the top left hand corner - the first four rows - engineering, clinical medicine, chemistry and physics under USA, China, UK, Australia and Japan - are considerably larger than the rest of the numbers.

Similar data are available with us for other Asian countries. Arunachalam has already published such data for India and China.(Ref. 7)


Conclusion


As pointed out by Hicks and Katz, at the macro level (involving all fields and all countries) international collaboration has been increasing at a constant rate since the mid-1960s in both the USA and UK.(Ref. 1, 12) The trend has been the same in other advanced countries. By and large the growth in international scientific collaboration, as reflected by jointly authored papers, paralleled the growth in the number of airline passengers and the number of international telephone calls in these countries. As it happens, in all walks of life there is a time lag before the less advanced countries catch up with such trends. Based on data from SCI 1991, Arunachalam et al.(Ref. 8) pointed out that South-South collaboration was meager. By 1998 the situation has improved considerably for collaboration involving the three Asian giants and other Asian countries. In particular, China, South Korea, Taiwan and Singapore have become much more international. China, in particular, seems to be emerging as a regional focal point.


Acknowledgement


We thank Mr R Chendil Kumar of M S Swaminathan Research Foundation, Chennai, for assistance in data processing. We are grateful to Dr Hildrun Kretschmer for making it possible for one of us to present this paper at the Second Berlin Workshop on Scientometrics and Informetrics, 1-4 September 2000.

References

1. Hicks, D. and Katz, J.S., Science and Public Policy, 1996, 23 (No. 1), 39-44.
2. Okubo, Y. and Miquel, J-F., International cooperation in basic science. In: P. Weingart, R. Sehringer and M. Winterhager (eds), Representations of Science and Technology (University of Bielefeld, Centre for Science Studies, 1990), 124-143.
3. European Commission, Second European Report on Science & Technology
Indicators, 1997. Brussels - Luxembourg, 1997.
4. National Science Board, Science & Engineering Indicators - 2000. Arlington, VA:
National Science Foundation, 2000 (NSB-00-1).
5. Arunachalam, S. Curr. Sci., 1992, 63: 56-58.
6. Science and Technology Policy: Review and Outlook 1991 (OECD, Paris, 1992). See chapter 3, pp. 69-80.
7. Arunachalam, S., International collaboration in science: The case of India and China, in The Web of Knowledge: A Festschrift in Honor of Eugene Garfield, edited by B. Cronin and H. B. Atkins. Medford, NJ: Information Today Inc. & The American Society for Information Science. September 2000.
8. Arunachalam, S., Srinivasan, R. and Raman, V., Scientometrics, 1994, 30: 7-22.
9. Arunachalam, S. and Jinandra Doss, M., J. Inf. Sci., 2000, 26:39-49.
10. Frame, J. D. and Carpenter, M. P., Social Studies of Science, 1979, 9: 481-497.
11. Miquel, J-F., Indicators to measure internationalization of science: The method and a few results. Report to the Office of Science and Technology, Washington, DC.
12. Katz, J. S. and Hicks, D., Policy processes, Proc. Fourth Int. Conf. on S&T Indicators, Antwerp, Belgium, 1995, pp. 75-79.

Back to Papers ListBack to Papers List
Top