Dynamics of behaviour and information diffusion on complex networks: analytical and empirical perspectives

Networks are ubiquitous in the world around us. Any system of interacting objects can be conveniently represented as a network, allowing for mathematical interrogation of its properties. These systems range from news and social networks (such as Twitter, Facebook, Google+ or school interaction networks) to physical systems (such as gene interactions networks, gas or electric networks). In this thesis, we investigate the effect of connectivity patterns of individuals in social networks on the spreading of behaviour and information from, analytical and empirical perspectives. The connectivity patterns that society are organised into mediate the individuals’ behaviour and in turn the behaviour of the network as a whole. For example, the adoption of a behaviour and the spreading of information is shaped by these same connectivity patterns. Providing a rigorous mathematical understanding of such diffusion process is of key importance for a broad range of domains from the social sciences, epidemiology to commercial interests. We present an analytically tractable model for the spreading of the adoption of behaviours on clustered clique-type networks. We successfully address the shortfalls of previous models, account for clustering, and provide an analytical validation of experimental results for the diffusion of behaviour on clustered networks. Second, we presented industry work modelling subscriber retention for a telecommunications company, where we provide a detailed understanding of subscriber behaviour and successfully predict non-renewal of subscriptions. We examined the relationship between community structure, user sentiment and ideological disposition on Twitter surrounding the Irish Marriage Referendum in 2015. We propose a novel approach to finding users of differing ideological dispositions. Lastly, we leverage the ideological communities to create a novel framework that enables us to examine how information (in the form of retweets) diffused between the differing ideological groups involved in the debate about the referendum.