MÉTRICAS INTERNAS ESCALA CRITERIO DE VALORACIÓN

As stated earlier, a positive correlation is seen between the number of nodes (and edges) in the network and the size of the giant component, within the context of these four engineering disciplines. However, looking from another perspective, just the existence of a large number of nodes (authors) in a network cannot be the sole reason for the formation of a giant component. For example, MECH has 809 nodes; yet, the largest component is just at 13.27% even after over three decades of activity. Even a very small network of just 48 researchers of COLLNET (Yin et al., 2006), a dedicated research forum of scientists studying scholarly collaboration networks, had a largest component possessing 32 nodes or 66.6% of the total network. Hence, just the presence of large number of nodes is no guarantee that a giant component would exist in such networks. It may be that scientific network possessing a large number of nodes, but nodes working separately in diverse sub- disciplines, would still keep the network fragmented for a long time. Engineering disciplines have dedicated sub-disciplines. For example, Mechanical engineering may have ‘complex mechanics’ and ‘micro-mechanical science’ as two separate divisions or sub-disciplines. In Universities, these sub-disciplines are sometimes enshrined as separate departments within the faculty. Such categories within a discipline can lead to fragmentation as researchers generally have favorable circumstances to collaborate with fellow researchers within their research divisions. One way to see faster formation of giant component is by fostering collaboration between these sub-disciplines. After all, it takes just one edge to bring two components or clusters of researchers together. Additionally, unlike random networks, collaboration in real-world networks, such as, co- authorship network, follows a certain pattern, also known as preferential attachment

(Newman, 2002). As such, some nodes attract connections by virtue of these nodes being already well connected or due to some other kind of assortative mixing (Newman, 2002).

There seems to be no particular cause for the formation of giant components. Although, rise in the number of research articles or increase in collaboration among researchers might play an important role, they cannot be standalone reasons for the formation of giant components. Rather, a variety of causes working in tandem may be responsible for the formation of giant components.

4.3 Research Question 3

Here the results and analysis of Research Question 3 which represents Case Study 3, are presented. RQ3 is restated here:

How do collaborative networks of Malaysia and Turkey, the two OIC nations, compare with each other in the field of ‘energy fuels’?

This research question has specific sub-objectives and sub-questions and I attempt to describe and answer each, one by one, as I progress with the description of results.

The Turkey dataset returned 2,150 authors who have published 1,658 articles in 79 journals. Citations per paper averaged 5.82. The Malaysia dataset returned 1,234 authors who published 658 articles in 69 journals. Malaysia’s output shows an incremental rise on a year-on-year basis. In 2009, Malaysia’s output was 169 articles; in 2010 it rose to 204 articles; and in 2011 it rose further to 285 articles. This increase also corresponds to Malaysia’s impetus in publishing in ISI-ranked journals in recent years. Unlike Turkey, Malaysia received more global citations (total number of citations to papers in WoS) for 2010’s published papers compared with 2009. This may be due to the increased number

124 of papers published in 2010. The average citation per paper stood at 5.65, which is very similar to Turkey. In Turkey, the number of country-based self-citations is much higher than in Malaysia. In Malaysia, only 1 in 6 citations came from papers written locally, whereas in Turkey this number is 1 local citation in every 3.82 citations. The bibliometric statistics of both countries are depicted in Table 4.3.1.

In both countries, public universities seem to perform better in research productivity. These institutions in Turkey, namely, Middle East Technical University (METU), Ege University, Istanbul Technical University, Gazi University, and Firat University were among the most productive in Turkey. In the same light, public universities in Malaysia, namely, the University of Malaya (UM), University Science Malaysia (USM), University Kebangsaan Malaysia (UKM), University Technology Malaysia (UTM), and University Putra Malaysia (UPM) garnered more than 65% of the total number of published papers. These five public universities in Malaysia are also designated as RUs, or ‘Research Universities’, and have received generous research grants from the Malaysian government (Abrizah & Wee, 2011). The research output of top universities in Malaysia contrasts sharply with that of Turkey, where the top five institutions garnered only 30% of the published papers.

Table 4.3.1: Bibliometric Statistics of the Turkey and Malaysia Datasets

Turkey Malaysia

Number of papers 1658 658

Number of authors 2150 1234

Mean citations per paper 5.82 5.65

Number of journals 79 69

Number of countries collaborated with 44 48

Single-author papers 427 16

Mean number of authors per paper 2.55 3.77

Mean number of papers per author 1.96 2.01

A significant percentage of 5-author (13%) and 6-author (6%) papers are found in the Malaysia dataset. This is not evident in the Turkey dataset.

Lotka (1926) investigated the frequency distribution of author productivity among chemists and physicists and found that the number of authors writing n articles is about 1/n2 of those writing one paper, and the proportion of all authors that make a one-paper contribution is about 60%. Since publishing his findings, Lotka’s measures are now established as Lotka’s Law of Scientific productivity (Talukdar, 2011). The author productivity fit using ‘Lotka’ software (Rousseau & Rousseau, 2000) found that the Turkey and Malaysia datasets fit Lotka’s Law with 𝛽 = 2.2858 and 2.326, respectively.