Loading...
On modelling and data science approaches towards understanding the dynamics of social diffusion and success within complex systems
Date
2021
Abstract
Complexity science has proven extremely successful in describing systems from a wide range of domains across the spectrum of society. Generally these understandings arise from considering the interaction of entities underlying the system using a combination of tools found within the far from disparate areas of mathematical modelling, network science, and data analysis. In this thesis we consider a range of such problems from three such perspectives. First, we consider the applicability of branching processes, a tool originally utilised in population dynamics, in describing social spreading phenomena. This involves developing models which describe the spreading of online content or ‘memes’ across networked systems and also the dynamics of individual cascades. Furthermore we develop mathematically tractable frameworks for providing pre dictions regarding the future popularity of content. We then study and extend some classical models used in describing how some pieces of content become ex tremely popular while the opposite behaviour occurs for the majority of others. Second, we proceed to conduct data-driven analyses into the realm of sporting contests. This results in studies regarding both fantasy and physical sports within which we consider both the reasons as to why some competitors become successful and also approaches used to quantify this success via tools from network science. Lastly, the directionality of interactions within a range of empirical systems are considered and its resulting effect upon dynamics which occur on the corresponding networks are studied by means on information theoretical approaches. We observe ubiquitous properties across the different domains suggesting a consistent hierarchical structure within complex systems.
Supervisor
Gleeson, James P.
Description
peer-reviewed
Publisher
Citation
Collections
Files
Loading...
OBrien_2021_Modelling.pdf
Adobe PDF, 8.15 MB