Structure of research collaboration networks – case studies on Malaysia / Sameer Kumar

• Main Author: Kumar, Sameer
• Format: Thesis
• Published: 2013
• Subjects: Q Science (General)
• Online Access: http://studentsrepo.um.edu.my/4370/1/PhD_Thesis_%2D_SAMEER_KUMAR_%2D_QHA100003.pdf; http://studentsrepo.um.edu.my/4370/

This study examines research collaborations in Malaysia from the perspective of networks. A research collaboration network is formed by connecting two researchers who have co-authored a research paper. Through three essay-based case studies, each representing a research question, topological properties of networks are investigated using social network analysis. In addition, some of the long-standing questions in research collaborations are answered, and an effective co-authorship strategy and a method to detect academic communities are proposed. The first case study investigates research collaborations in the field of business and management in Malaysia. After manually disambiguating the authors, the network of 285 business and management researchers at the individual, institutional and international levels were examined. The study found that the popularity of researchers and the strength and diversity of their ties with other researchers had significant effects on their research performance. Furthermore, geographical proximity mattered in intra-national collaborations. Surprisingly, Malaysia has had relatively little collaboration with other ASEAN nations, although it is a prominent member and has an important agenda of educational cooperation with its member states. Internationally co-authored articles have been cited almost three times more frequently than locally co-authored articles. Based on these results, a strategy for co-authorship is suggested. In the second case study, the size of the giant component of co-authorship networks was investigated in the four prominent engineering disciplines: electrical and electronics (EEE), chemical (CHEM), civil (CIVIL), and mechanical (MECH), involving 3675 scholarly articles, in which at least one of the researchers per article had a Malaysian address. Results revealed that well-formed giant components (size >50% of all nodes) were already present in EEE and CHEM disciplines, whereas they were at an undeveloped iii stage in both CIVIL and MECH. However, those with larger giant components also had larger degree of separation (geodesic distance) between the nodes. Density of the nodes was negatively correlated with the size of the giant component. After the mid-1990s, both CHEM and EEE had a faster production of articles than the other two disciplines, corresponding with their well-formed giant components. The third case study, collaborative patterns of Malaysia, was compared with another OIC country, Turkey, in the fast-developing field of Energy Fuels. The popularity, position and prestige of the authors in the network, as determined through centrality measures, had statistically significant effect on research performance. However, these measures were far more correlated with the research performance of the authors in the Malaysia network than in the Turkey network. Authors’ degree (‘deg-core’) was applied to reach to the core of network, which in contrast to standard K-Core method, was found to capture more productive authors. A method to detect academic communities of productive authors by extracting motifs (large cliques) from the network is suggested. Finally, the cognitive structures of both countries using a 2-mode network representing research focus areas (RFAs) and prominent authors working in these RFAs were visualized.