A methodology for multilevel analysis of scientific collaboration networks : mapping current computer science research in New Zealand
Martin, Bernd H.
Citation:Martin, B.H. (2013). A methodology for multilevel analysis of scientific collaboration networks : mapping current computer science research in New Zealand. An unpublished thesis submitted in partial fulfilment of the requirements for the Degree of Master of Computing.
Permanent link to Research Bank record:http://hdl.handle.net/10652/2430
This research scientifically analysed the evolving Complex Network structures of the New Zealand Computer Science research community upon multiple levels (Macro, Meso, Micro, Topics). Methodological approaches utilised interdisciplinary techniques comprised of data mining, social network analyses, scientometrics and data visualisation. The research sought to identify communities, highly influential nodes, research institutions, and their collaborative patterns over the last 5 years. Network metrics revealed insights into the structure of the networks. Collaboration networks were generated using a variety of layout algorithms then visually presented in the form of knowledge maps. Furthermore, Word Co-occurrence networks of terms from both the Titles field and Keywords field were constructed and analysed to reveal topic trends and bursts. The mapping of recent New Zealand Computer Science research developments was accomplished by using Alluvial diagrams. The change of streams over the time period highlights the nature of, and evolving relations within and amongst topics. The visual results of this research provide a natural way to reveal information. To my knowledge, this is the most comprehensive multilevel study of a specific domain (Computer Science) conducted within New Zealand, to date. The applied methods are transferable to other domains and interdisciplinary endeavours. A real world application of the applied methodology could be an enhancement of the existing interdisciplinary portal (www.nzresearch.org.nz/) with the application of multilevel analysis methods. This could enable collaboration and discovery among scientists across all disciplines. An interactive multimedia presentation of this research including high definition maps and a 3D demonstration of the Topic network can be found at: [tba] It accompanies, supports the findings of, and enhances this written research.