Geographical Space
ZSÓFIA VIKTÓRIA VIDA 1
3. Data and methods
For testing our concept, the separation of social and pure cognitive distance within cognitive distance, we analyzed WoS records of two fields with cited references between 2010–2014 which contained at least one Hungarian author. The examined two fields are an SSH field, Economics, and a Science field, Physical Geography. This choice gives the chance to compare our findings between SSH and Science. To determine fields, the Web of Science Categories (WCs) were used. We collected papers not only from a single WC but from WCs which are in strong relation with Economics or Physical Geography. Thus, we get a similar sized dataset in the two fields (Table 1). To obtain the groups we clustered the WCs via journal and WC co-occurrence matrices and used hierarchical clustering (Vida, 2016). The chosen WC groups are:
Economics:
Agricultural Economics & Policy;
Business, Finance;
Economics.
Physical Geography:
Geography, Physical;
Geosciences, Multidisciplinary;
Imaging Science & Photographic Technology;
Remote Sensing;
Engineering, Geological.
Table 1. Size of dataset – Hungarian articles between 2010–2014 in two fields
From now on, Economics means the group of Economics and Physical Geography means the group of Physical Geography.
We analyzed those records which contained cited references (CR), a necessary thing for author bibliographic coupling. CRs were given in higher rate in Physical Geography than in Economics.
Determining social and cognitive distances we used similarity matrices. The higher the cell content was, the lower the distance was between two authors.
We set up an adjacency matrix via author bibliographic coupling and we determined the similarity between authors with Salton’s Cosine similarity (Hamers et al., 1989; Nguyen, Bai, 2010). The higher the cell content was in the similarity matrix, the more common references were there between authors. The values of the cells were between 0 and 1. This was the entire cognitive similarity matrix which contained both components (the bottom of Figure 2).
As we saw in the bottom right part of Figure 2, the social component derives from co-authorship. Thus, in determining social distance, we described the co-authorship with similarity matrix using Salton’s Cosine similarity.
Determining pure cognitive distance, first we obtained the pure cognitive similarity matrix by subtracting the social component similarity matrix from the entire cognitive similarity matrix. Then we used absolute value of the results.
All the similarity matrices can determine a network, where the authors are the nodes and the similarity values are the edges. So, we created three weighted and undirected networks from the three similarity matrices. Then we compared them via the attributes of the networks and via Quadratic Assignment Procedure (QAP) correlation. Finally, we projected the networks on geographical maps. For the calculation we used R (R Core Team, 2015; Csardi, Nepusz, 2006; Meyer, Buchta, 2015).
4. Findings
The created similarity networks determined weighted and undirected networks, where the nodes are the authors and the edges are the similarity values. In all the networks the nodes are the same.
In Table 2 we see the main attributes of the networks. In the cognitive networks we can observe a higher number of relations: these relations came from the similarity of cited references so these relations were artificial connections, whereas the social networks contained real connections because these derived from co-authorships. The density values showed similar attributes with the social networks being the least dense. However, all density values were very low.
In the corpus of the present study the ratio of co-authored papers in Physical Geography and Economics was respectively 89% and 61%, whereas the average number of co-authors of a paper was 6 and 3. In the case of all the three distance types there were much more edges in Physical Geography.
Table 2. The main attributes of the networks
To validate the model, that is, to prove the difference between social and pure cognitive components, we used QAP correlation.
In both scientific fields there was a strong connection between the entire cognitive network and the social network, whereas pure cognitive networks had weak connections with both entire cognitive and social networks. After filtering social relations, the pure cognitive network showed a weak relation with the entire cognitive network in both cases (Table 3). The QAP correlation showed a more powerful presence of the social component within entire cognitive distance. From this point of view, the two examined science fields presented a similar picture, even if the social component was a bit stronger in the case of Physical Geography.
This might be explained with the bigger frequency of co-authorship in this field.
To investigate the appearance of social and cognitive distance in geographical space we projected the social and cognitive networks on geographical maps, thus connecting inner and outer spaces. Instead of authors the relations were presented between cities, based on the affiliation of authors. We did not aggregate the network: the lines representing edges were projected one above the other, so the presence of a line shows at least one established connection between authors of two different cities.
Table 3. Results of the QAP correlation
Economics Physical Geography
# nodes 704 2294
# edges 5145 67282
density 2,0792 2,5582
# nodes 704 2294
# edges 1347 16351
density 0,5443 0,6217
# nodes 704 2294
# edges 4874 61157
density 1,9696 2,3253
Entire cognitive
network Social network
Pure cognitive
network
The relations of social and cognitive networks covered almost the whole world.
In this study we show only relationships within Europe, since without the connections with the United States the strongest relations were here.
In Figure 3 we see the edges with weights above median of social and pure cognitive networks within Europe in the fields of Economics. The clustering effect of the networks accounts for relations between two non-Hungarian cities.
Figure 3. Edges of social and pure cognitive networks with weights above median within Europe in the field of Economics
The more important social connections covered almost all countries of Europe.
Our country’s main collaborating partners were Germany, the United Kingdom, Belgium, the Netherlands and the neighbors of Hungary, e.g. Romania. In the aspect of cognitive relations we got a west-oriented connection. This means that Hungarian and Western European researchers used similar literature.
In the case of pure cognitive networks this orientation towards West was present only in the strongest 10 percent of the relations in the fields of Physical Geography (Figure 4).
Figure 4. The strongest 10% of relations in the pure cognitive network in the field of Physical Geography within Europe
5. Conclusion
In this study we investigated scientific collaborations from the aspect of the distance between the actors which has an important role in their formation.
According to the literature we distinguished three types: geographical, social and cognitive distance. We analysed the author level where these relations are established.
The social distance was studied via co-authorship networks and the cognitive distance was analysed via BC on the author level, with author bibliographic coupling. During the projections to the author level we kept all the authors of a publication. Using author bibliographic coupling we found two reasons for the similarity of references between two authors:
1. the two authors are co-authors, so the co-authored papers’ cited references appear for both authors (social component of cognitive distance);
2. the two authors are not co-authors but their research area is close to each other so the used references are similar (pure cognitive component of cognitive distance).
In this study we separated the social component of cognitive distance from the pure cognitive distance.
We used QAP correlation to prove the separation of the two components. Our findings proved that pure cognitive distance between authors can show a different picture from entire cognitive distance because of the filtering of the social factor.
With the help of pure cognitive distance, we can sign potential future collabora-tions.
During the projection of the social and cognitive dimensions into geographical space, in both analysed science fields, the social and cognitive relationships of Hungary were realized on one hand with geographically near countries, on the other hand with countries with a high publication emission and high citation indicators at international level. A typically West-oriented attitude was clearly visible in the case of pure cognitive connections rather than in the case of social connections. The two scientific fields showed similar patterns although the more frequent collaboration caused a stronger social component in Physical Geography.
References
ACOSTA, M., CORONADO, D., FERRÁNDIZ, E., LEÓN, M. D. 2011. Factors affecting inter-regional academic scientific collaboration within Europe: The role of economic distance. In:
Scientometrics. Vol. 87, No. 1, pp. 63–74.
BALDI, S. 1998. Normative versus social-constructivist processes in the allocation of citations: A network-analytic model. In: American Sociological Review. Vol. 63, No. 6, pp. 829–846.
BEAVER, D. D. 2001. Reflections on scientific collaboration (and its study): past, present, and future. In: Scientometrics. Vol. 52, No. 3, pp. 365–377.
BOSCHMA, R. 2005. Proximity and innovation: a critical assessment. In: Regional studies. Vol.
39, No. 1, pp. 61–74.
BOYACK, K. W., KLAVANS, R., BÖRNER, K. 2005. Mapping the backbone of science. In:
Scientometrics. In: 64, No. 3, pp. 351–374.
BOYACK, K. W., KLAVANS, R. 2010. Co-citation analysis, bibliographic coupling, and direct citation: Which citation approach represents the research front most accurately? In: Journal of the Association for Information Science and Technology. Vol. 61, No. 12, pp. 2389–2404.
CSARDI, G., NEPUSZ, T. 2006. The igraph software package for complex network research. In:
InterJournal, Complex Systems. Vol. 1695, No. 5, pp. 1–9.
DE SOLLA PRICE, D. J., BEAVER, D. 1966. Collaboration in an invisible college. In: American Psychologist. Vol. 21, No. 11, pp. 1011–1018.
DE SOLLA PRICE, D. 1979. Kis tudomány – nagy tudomány. Budapest: Akadémiai Kiadó.
FRENKEN, K., HARDEMAN, S., HOEKMAN, J. 2009. Spatial scientometrics: Towards a cumulative research program. In: Journal of Informetrics. Vol. 3, No. 3, pp. 222–232.
GLÄNZEL, W., SCHUBERT, A. 2004. Analysing scientific networks through co-authorship. In:
MOED, H. F., GLÄNZEL, W., SCHMOCH, U. (rds.) Handbook of quantitative science and technology research. Dordrecht: Springer. pp. 257–276.
HAMERS, L., HEMERYCK, Y., HERWEYERS, G., JANSSEN, M., KETERS, H., ROUSSEAU, R., VANHOUTTE, A. 1989. Similarity measures in scientometric research: the Jaccard index versus Salton's cosine formula. In: Information Processing & Management. Vol. 25, No. 3, pp.
315–318.
HOEKMAN, J., FRENKEN, K., TIJSSEN, R. J. 2010. Research collaboration at a distance:
Changing spatial patterns of scientific collaboration within Europe. In: Research Policy. Vol.
39, No. 5, pp. 662–673.
IKPAAHINDI, L. [SCHUBERT, A. ford.] 1986. An overview of bibliometrics: its measurements, laws and their applications [A bibliometria és a tudománymetria mérési módszerei, törvényei és alkalmazásai]. In: Kutatás-fejlesztés, Tudományszervezési tájékoztató. Vol. 26 No. 3–4, pp. 279 –294.
JARNEVING, B. 2007. Bibliographic coupling and its application to research-front and other core documents. In: Journal of Informetrics. Vol. 1, Vol. 4, pp. 287–307.
KATZ, J. S. 1994. Geographical proximity and scientific collaboration. In: Scientometrics. Vol. 31, pp. 1, pp. 31–43.
KATZ, J. S., MARTIN, B. R. 1997. What is research collaboration? In: Research policy. Vol. 26, No. 1, pp. 1–18.
KESSLER, M. M. 1963. Bibliographic coupling between scientific papers. In: Journal of the Association for Information Science and Technology. Vol. 14, 1, pp. 10–25.
KESSLER, M. M. 1965. Comparison of the results of bibliographic coupling and analytic subject indexing. In: American Documentation. Vol. 16, No. 3, pp. 223–233.
LUUKKONEN, T., TIJSSEN, R., PERSSON, O., SIVERTSEN, G. 1993. The measurement of international scientific collaboration. In: Scientometrics. Vol. 28, No. 1, pp. 15–36.
MARTYN, J. 1964. Bibliographic coupling. In: Journal of Documentation. Vol. 20, No. 4, pp. 236–
236.
MELIN, G., PERSSON, O. 1996. Studying research collaboration using co-authorships. In:
Scientometrics. Vol. 36, No. 3, pp. 363–377.
MEYER, D., BUCHTA, C. 2015. proxy: Distance and Similarity Measures. R package version 0.4-15. https://CRAN.R-project.org/package=proxy
NARIN, F., STEVENS, K., WHITLOW, E. 1991. Scientific co-operation in Europe and the citation of multinationally authored papers. In: Scientometrics. Vol. 21, No. 3, pp. 313–323.
NGUYEN, H. V., BAI, L. 2010. Cosine similarity metric learning for face verification. In: Computer Vision–ACCV 2010. Springer Berlin Heidelberg. pp. 709–720.
PAIER, M., SCHERNGELL, T. 2011. Determinants of collaboration in European R&D networks:
empirical evidence from a discrete choice model. In: Industry and Innovation. Vol. 18, No. 1, pp. 89–104.
R Core Team 2015. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/
RADICCHI, F., FORTUNATO, S., MARKINES, B., VESPIGNANI, A. 2009. Diffusion of scientific credits and the ranking of scientists. In: Physical Review E. Vol. 80, No. 5, 056103.
SMALL, H. 1973. Co-citation in the scientific literature: A new measure of the relationship between two documents. In: Journal of the Association for Information Science and Technology. Vol. 24, No. 4, pp. 265–269.
SONNENWALD, D. H. 2007. Scientific collaboration. In: CRONIN, B. (ed.) Annual Review of Information Science and Technology, Vol. 41 (pp. 643–681), Medford (NJ): Information Today.
TAGLIACOZZO, R. 1967. Citations and citation indexes: A review. Methods of Information in Medicine. Vol. 6, No. 3, pp. 136–142.
VIDA, Z. V. 2016. “Scientific Collaborations” at the Level of Countries – A Case Study: A “Hard”
Physical Geography and a “Soft” Economics Science Field. In: Deturope. Vol. 8, No. 3, pp.
224–246.
WHITE, H. D., WELLMAN, B., NAZER, N. 2004. Does citation reflect social structure?:
Longitudinal evidence from the “Globenet” interdisciplinary research group. In: Journal of the American Society for Information Science and Technology. Vol. 55, No. 2, pp. 111–126.
WUYTS, S., COLOMBO, M. G., DUTTA, S., NOOTEBOOM, B. 2005. Empirical tests of optimal cognitive distance. In: Journal of Economic Behavior & Organization. Vol. 58, No, 2, pp. 277–
302.
YAN, E., DING, Y. 2012. Scholarly network similarities: How bibliographic coupling networks, citation networks, cocitation networks, topical networks, coauthorship networks, and coword networks relate to each other. In: Journal of the American Society for Information Science and Technology. Vol. 63, No. 7, pp. 1313–1326.