University networks in the context of their academic excellence and openness: A comparative study of leading Czech and German universities

Vladimir M. Moskovkin
Belgorod State University, Russia, Belgorod, 85 Pobeda St., Russia. Tel.: 810-4722-30-13-00, (ext. 27-70). E-mail: moskovkin (at) bsu.edu.ru

Jason K. Fraser
Belgorod State University, Belgorod, 85 Pobeda St., Russia. Tel.: 810-4722-30-13-00, (ext. 27-70). E-mail: goldanddiamond (at) live.com

Maria V. Moskovkina
International Slavonic University, Kharkov, Otakar Yarosh St., 9-A, Ukraine. Tel. / Fax +38 (057) 340-03-65. Email: info (at) msu.kharkov.ua

Received June 30, 2012; Accepted January 25, 2013

Abstract

A simple methodology of multi-dimensional vector analysis for the comparison of the academic performance and the openness of university networks of the identical dimension was developed, which is illustrated by the example of the leading universities in the Czech Republic and Germany. In order to make this comparison, proximity measures were introduced with an arbitrary normalized vector of indicators to standard unit vectors that can be found with the aid of the normalized Euclidean distance. As an example of the indicators of academic performance and openness, the number of universities included in known global university rankings and having open access repositories and English versions of the university's website, and participating in Berlin Declaration of Open Access, SINAPSE platform, European University Association and the Magna Charter is considered.

Keywords

Multi-dimensional vector analysis; University networks; Czech universities; German universities; University ranking; Open access; Institutional repository

Introduction

In the current conditions of globalization, competition among universities increases and plays an important role. Universities compete among themselves at global level for the best students and researchers, for acquisition of contracts on Research and Development from the governments, businesses and public funds.

The capability of the university to compete at global level with other universities is known as its global academic competitiveness. For its quantitative assessment, different university ratings are offered that allow to compare different universities, based on the degree of academic excellence.

Apart from the indicators of academic excellence, which are taken into consideration in global university rankings, we can also consider the openness indicators of the universities, which also promotes growth of their global academic competitiveness; that is, OA (open access) repositories, English websites, and membership of the universities in other academic and university unions that allows to exchange knowledge and best university practice. If desired, indicators of openness could be introduced into the system of indicators of global university rankings.

Besides, comparing among themselves of separate universities on the basis of academic excellence and openness indicators, that determines their global academic competitiveness in the form of aggregated university ratings, it is possible to compare country sets of universities based on the same indicators. It is obvious that for such comparisons, country sets have to be in the same dimensions. In the capacity of such sets, we can select TOP-N universities in every country of research, on any wide university rating. Where N is the number of universities under study.

The problem of comparative analysis of university networks with the same dimensions (N) in the context of their academic excellence and openness, is based on the calculation of the occurrence of universities included in TOP-N, in various global university rankings and university and academic associations, calculation of existence of platforms of open access to them (OA-repositories, English websites) and also introductions of a measure of proximity to some ideal vector corresponding to the maximum number of the above-mentioned occurrences and the presence of platforms of open access.

At present, the formulation of this kind of problem is not available. This can be proved by means of testing relevant scientific terms on Google Scholar and the content analysis of the received responses. Usually, when dealing with comparative analysis of different networks, the emphasis is put on the interaction of network nodes, instead of occurrence nodes into set of any features.

Cooper and Barahon (2011) proposed a new measure of similarity between nodes in different networks. Here the matrix of similarity related to the distance between feature vectors that contain input and output paths of all the segments for each node.

A research by Varga and Parag (2009), performed within FP7, was done on the basis of classical works in the area of building a concept of a National Innovation System (Lundvall, 1992; Nelson, 1993) and the following conclusion can be made. The efficiency of research networks in producing new knowledge can be approached by three features: the number of actors involved in the system (the size of the network), the knowledge those actors have accumulated and the intensity of knowledge-related interactions among the actors during knowledge creation. If we consider the actors in the system or network as universities, then in analogy with the above-mentioned paper, we can conclude that the size of the university network, the knowledge level of the universities and the intensity of knowledge related linkage characterize the university network connection quality. If the paper (Varga & Parag, 2009) looks at the international co-publication networks of different sizes that are generated by the University of Peos academic unit, we will examine and compare Czech and German university networks of the same size and abstracting from interactions in these networks. We study the reputation (accumulated knowledge) of the members of these networks (separate universities).

In the research by Larivière, Gingras and Archambault (2006), which was on the basis of UCINET (Borgatti, Everett & Freeman, 2002) and NETDRAW (Borgatti, 2002) a network analysis software programs was built by the Canadian inter-institutional Collaborative networks (English-speaking and French-speaking universities with 30 or more joint publications, 1980-2002) on the social sciences (SS) and natural sciences and engineering (NSE). It was shown that almost all articles on the NSE are jointly published, compared to two thirds on the SS and about 10% in the humanities, as well as, the bibliometrics mapping of collaborative networks gives a very good idea of overall trends in collaboration and highlight the gaps between the humanities, SS and NSE. Comparative analysis of inter-institutional networks shows that SS are probably nearer to the NSE than to the humanities.

The study by Cromwell et al. (2011) examines an instrument that contains five online modules, that latter one was Research Portfolio: a tool to examine the metrics of research productivity and building research networks for NIH grant acquisition. This tool is used by six institutions with Clinical and Translational Science Awards (CTSA) for building Biometrical Resource Ontology (BRO) search terms. A distribution matrix for BRO research terms for six universities that are included in CTSA was built.

In general, the existing approaches used in formalized analysis for the comparison of networks are based on graph and matrix theory and network planning, in which the interaction between network nodes play a significant role. At the same time, there are no articles available that compare networks with the same dimensions in relation to their occurring nodes (in current case universities) in space of any features.

Materials and Methods

When formulating any scientific problem, it is necessary to understand the degree and level of scrutiny of this issue. To do this, we need to choose adequate English terms that are to be included in the conceptual apparatus of the given task. In the capacity of such terms were selected: university network, university networks, university networking, comparison of networks, research network, research networks, research networking, comparative analysis of networks, and mathematical analysis of networks. Such terms will be tested with the help of Google Scholar in advanced research mode in the line "with exact phrase" for two cases: "anywhere in the article" (two options: include citations; at least summaries) and "in the title of the article" (same option). Such experiments allows to prove or refute, the hypothesis done in the Introduction part about the absence of formulation of the problem on comparative analysis of university networks with the same dimensions in the context of their academic excellence and openness.

We will consider a set of networks in the amount of M of the same dimension N, where N is the amount of network nodes. For such networks, we introduce n - dimensional vector of indicators (features): = (V₁, V₂, …, V_i …, V_n), as well as normalized vector = ( ₁, ₂, …, _i…, _n), where _i= V_i/N. We assume that V_i satisfies the inequality 0 ≤ V_i ≤ N, which implies the inequality 0 ≤ _i ≤ 1. The proximity of an arbitrary vector to the standard unit vector that we find with the aid of the normalized Euclidean distance:

where 0 ≤ d ≤ 1.
The smaller d is, the closer the is to the standard unit vector. Thus we can rank all M networks with the same dimensions depending on their proximity to the standard unit vector in n - dimensional feature space.

We present an example of this simple methodology for two university networks, the Czech Republic and Germany, consisting of 45 universities (M = 2, N = 45). This is the number of universities which correspond to the number of ranked universities in Czech Republic with aid of the Webometric ranking in July 2010 (www.webometrics.info).

For a comparative analysis of the leading Czech and German university networks with the same dimensions (N = 45) we introduced the following system of indicators: the quantity of universities in the Top-1000 World Web Rank, Top-1000 Scholar Rank, Top-200 British Times Rank, Top-500 Shanghai Rank, Top-500 Taiwan Rank, the amount of university OA-repositories and English versions of university sites, participants in the Berlin Declaration of Open Access, SINAPSE platform, European University Association and the Magna Charter (n =11). Note that proposed academic and openness indicators, along with others (living conditions and safety on college and university campuses, scholarship and grant support, cost of living in cities where universities are located, etc.), can be used in constructing a simulation expert system of choosing universities for training and research (Moskovkin, 2009).

When we talk about academic excellence, we mean powerful research universities, the competence that is reflected in their publication activity and the citation of their scholars' articles as well as the volume and quality of training of their scientific personnel (PhDs).

To the indicators of academic excellence, we will assign Spanish (Webometric), British, Shanghai and Taiwan rankings as well as Scholar index of the Spanish Webometric ranking.

The remaining indicators characterize, to a greater extent, the openness of universities, which is connected to the integration of these universities into the international movement of open access to scientific knowledge and higher education. From the point of view of Varga and Parag (2009) we can conclude that separate universities are included into the network of partner universities and academic establishments for which we can offer a measure of network connection quality in analogy with the above-mentioned article. But that is a task to be examined in another research.

The above suggested eleven indicators in varying degrees are responsible for the academic excellence and the openness of the universities.

Results and Discussions

Above-mentioned terms were tested with the help of Google Scholar and the results are presented in Table 1.

The most relevant publications were found in the publication clusters generated by the terms "comparison networks" and "research networks". This consists of the following works (Cooper & Barahon, 2011; Varga & Parag, 2009).

In the "comparative analysis of networks" cluster, we found an article by Larivière, Gingras and Archambault (2006). In the "research networking" cluster, we found an article by Cromwell et al. (2011). All the above-mentioned articles are presented in the Introduction. In the remaining publication clusters, we found no articles that are related to the formalized quantitative analysis of university networks.

Now we will consider comparative analysis of Czech and Germany university networks, consisting of 45 universities, on the basis of 11 indicators that are selected for the academic excellence and openness of the universities.

All values of the quantitative and qualitative indicators for the Czech universities are listed in Table 2 and 3, and for universities in Germany in Table 4 and 5. Universities in these tables are ranked in descending order according to the webometric rankings (July 2010).

For the Czech universities only Charles University had Shanghai and Taiwan ranks that correlated with the Scholar rank indicator. According to the latest indicators, the University of Masaryk has very high World rank which can be connected to their large exemplary collection of scientific papers.

But not all publications belong to the scholars of this university. For example, such situations occur when web representations of articles in scientific journals are published on the basis of the university. Note that the placement of other scholar's articles in the University Open Access (OA)-repository is considered as bad practice, which can be penalized by the Spanish Cybermetric Laboratory when they calculate the Webometric ranking of universities. There were only two of these OA-repositories recorded in the universities of the Czech Republic at the end of September 2010 - at the Technical University of Ostrava and the University of Pardubice. The first university raises questions of inconsistencies in the amount of documents placed in the OA-repository with Scholar Rank value. Our inquiry on Google Scholar 06.10.2010, with the help of the operator site: URL-address showed that the Technical University of Ostrava had 1570 documents on their site. Consequently, Google Scholar did not index the bulk of the university's OA-repository documents (the total number of documents at the end of September 2010 amounted to 46 339 (Table 2).

If a good software is used to create the OA-repository (e.g., DSpace, Eprint, etc.), and if the collection of documents and bibliographic description (metadata) is properly done, then Google Scholar is very fast at indexing it, thanks to the OAI-PMH interface (Open Archives Initiative Protocol for Metadata Harvesting).

The University of Pardubice has the best Scholar Rank indicator with a smaller number of documents in its OA-repository, when compared to the Technical University of Ostrava. In this case, Google Scholar has indexed on the 06.10.2010, a total of 14 000 documents which is comparable with the total number of documents (17 823) located in OA-repository of the University of Pardubice at the end of September 2010 (Table 1).

Almost all Czech universities, except for five, have English-language versions of their sites (Table 2). The Metropolitan University of Prague has the greatest (openness) transparency to the outside world, and in addition to having a Czech and English version of the site, it also has Russian, German and French versions. But their poor quality gives them a low university webometric rating (6400^th place in the world). We are interested in the Czech universities that have Ukrainian and Russian language versions of their sites, which indicate their intent to attract students from Ukraine and Russia and to collaborate with universities in these countries. The Ukrainian language version of the site has a Banking College in Prague, the Russian-speaking - Metropolitan University of Prague, Moravian College Olomouc, Hospitality (Gotelny) College in Prague and the Prague International University.

Table 3 shows the participation (September-October 2010) of Czech universities in the university-wide and academic communities. None of them expressed interest in signing the Berlin Declaration on Open Access to Knowledge in the Sciences and Humanities, which is reflected in the substantial absence of OA-repositories in the Czech universities and their low webometric ratings (Table 3). Only two Czech universities joined the European platform SINAPSE (Scientific Information for Policy Support in Europe): Karl and Masaryk Universities.

A relatively large number of Czech universities joined the European University Association (18 of 45), and about eight universities out of 45 joined the Magna Chapter (Bologna). The latter show that the relation to the Bologna Process in the Czech Republic is relatively modest, in contrast to the situation in Ukraine.

Table 4 is similar to Table 2 for Czech universities. Here we see that all the 45 German universities are in the TOP-500 Webometric ranking as well as in the TOP- 1000 according to Scholar Rank. The latter is due to the presence of OA-repositories in most German universities (only 11 out of 45 universities did not have OA-repositories). Several universities had two OA- repositories. The number of documents in these repositories do not always correlated with the Scholar Rank. Most universities in Germany are included in the TOP-500 Shanghai and Taiwan rankings, 13 universities out of 45 are in the TOP-200 rankings of British Times Rankings.

Practically all German universities as of October 2010 had English versions of their sites (43 of 45); four universities had French-speaking versions, and some universities had Chinese, Dutch, Turkish, Russian and Spanish version of their site (Table 5).

In contrast to Czech universities, many German universities are members of the European Universities Association (32 of 45) and are signatories of the Magna Charter (20 of 45).

Only 5 universities have joined the SINAPSE platform. Against the background of good integration of German universities to the international movement of open access to scientific knowledge (34 out of 45 universities had their own OA-repositories), a surprisingly low percentage of universities have signed the Berlin Declaration on Open Access to Knowledge in the Sciences and Humanities (3 of 45). Perhaps this is due to the fact that this declaration was not initiated by the university community, but by the academic community represented by the Max Planck Society (Table 5).

Thus, a comparison of two university networks of the Czech Republic and Germany (45 universities) of same dimension, shows a significant qualitative superiority of network of universities in Germany.

For carrying out calculations using formula (1), we have constructed Table 6 from Tables 2- 5.

On the basis of Table 6, the normalized vectors of Czech and German university network indicators are as follows: _Czech= (0.20, 0.24, 0, 0.02, 0.02, 0.04, 0.89, 0, 0.04, 0.40, 0.18); _Germany= (1, 1, 0.29, 0.73, 0.84, 0.76, 0.96, 0.07, 0.11, 0.71, 0.44).

Using formula 1 we obtain: d = 0.8542 for the Czech Republic, d= 0.4969 for Germany. Hence we see that _Germany is much closer to the standard unit vector than _Czech, and therefore, the network of German universities developed significantly better with regard to academic excellence and openness, than the comparative network of Czech universities in the same dimension.

Note that the top 20 Czech (Table 2) and German (Table 3) universities were used in the study by Moskovkin, Delux and Moskovkina (2012) for building university publication structures with the help of Google Scholar.

Conclusion

On the basis of a simple multi-dimensional vector analysis, we built a formalized quantitative procedure for the comparative analysis of the academic performance of universities and the openness of university networks of the same dimension as illustrated by the example of the leading universities in the Czech Republic and Germany. The suggested system indicators can be significantly expanded, for example, with the help of other global university ratings (Leiden, QS, URAP, SIR).

Comparison of university networks of the Czech Republic and Germany (45 universities) of same dimension, shows significant academic excellence and openness of network of universities in Germany. In order to make this comparison, proximity measures were introduced. This is represented by proximity of an arbitrary normalized vector of indicators to standard unit vectors that can be found with the aid of the normalized Euclidean distance.

As zero vector is separated from unit vector on d=1, then taking the distance for 100% from earlier calculations, we can see that the network of German universities is closer on 85.42% - 49.69% = 35.73% to the standard unit vector compared to the network of Czech Republic universities.

The proposed academic and openness indicators, along with others (living conditions and safety on college and university campuses, scholarship and grant support, cost of living in cities where universities are located, etc.), can be used in constructing a simulation expert system of choosing universities for training and research.

References

• Borgatti, S.P. (2002). NetDraw: Graph visualization software. Harvard: Analytic Technologies.
• Borgatti, S.P., Everett, M.G., & Freeman, L.C. (2002). Ucinet for windows: Software for social network analysis. Harvard: Analytic Technologies.
• Cooper, K., & Barahon, M. (2011). Role-similarity based comparison of directed networks. Retrieved October 10, 2012, from http://arxiv.org/abs/1103.5582
• Cromwell, J.B., Crowley, R.S., Borromeo, C., Singh, H., Whelan, N., & Becich, M.J. (2011). iBIOFind: Intelligent biomedical resource identification for enhancing scientific productivity and research networking. Paper resubmitted to BMC Bioinformatics.
• Larivière, V., Gingras, Y., & Archambault, E. (2006). Comparative analysis of networks of collaboration of Canadian researchers in the natural sciences, social sciences and the humanities. Scientometrics, 68(3), 519-533.
• Lundvall, B.A. (1992). National innovation systems: Towards a theory of innovation and interactive learning. London: Pinter.
• Moskovkin, V.M. (2009). The potential of using the Google Scholar search engine for estimating the publication activities of universities. Scientific and Technical Information Processing, 36(4), 198-202.
• Moskovkin, V.M., Delux, T., & Moskovkina, M.V. (2012). Comparative analysis of university publication activity by Google Scholar (on example of leading Czech and Germany universities). Cybermetrics, 16(1), paper 2. Retrieved October 10, 2012, from http://cybermetrics.cindoc.csic.es/articles/v16i1p2.html
• Nelson, R. (1993). National innovation systems. A comparative analysis. New York/Oxford: Oxford University Press.
• Varga, A., & Parag, A. (2009). Academic knowledge transfers and the structure of international research networks. Working Paper IAREG 1.3d. http://www.iareg.org/fileadmin/iareg/media/papers/WP1_03d.pdf

Bibliographic information of this paper for citing:

Moskovkin, Vladimir M., Fraser, Jason K., & Moskovkina, Maria V. (2013).   "University networks in the context of their academic excellence and openness: A comparative study of leading Czech and German universities."   Webology, 10(1), Article 107. Available at: http://www.webology.org/2013/v10n1/a107.html