Comprehensive Modelling of Influence Diffusion in Complex Social Networks, an Agent-based Perspective
Influence diffusion modelling, analysis and applications draw tremendous attention to both researchers and practitioners since many organisations attempt to utilise its power to achieve business or political goals. A great many research works have been dedicated to the exploration of maximizing the spread of a particular influence in complex networks, e.g., social networks. However, influence appears to be a hybrid and complex effect caused by numerous factors, such as friendship affiliation, preferences, common communities, etc. Moreover, due to the sophisticated and dynamic environment where influences reside in, modelling influence diffusion in complex networks becomes a very challenging topic.
In this thesis, agent-based approaches and multi-agent systems have been employed to model the influence propagation in complex systems. In other words, the perspective of exploring the spread of influence diffusion stands at a microscopic level, where the dissemination of influences is driven by individuals' personalised traits and behaviours.
First, the thesis elaborates the hybrid effects of influence and systematically presents a generic architecture of modelling influences from an agent-based perspective. Second, based on the proposed framework, we further investigate the agent-based approach with stigmergic interactions, to address the influence maximisation problem in a dynamic and complex environment. Third, driven by the business needs for long-term marketing, the generic agent-based model has been extended by incorporating the capabilities for maintaining long-lasting influences. Last but not least, by considering the coexistence of multiple influences, the agent-based model has been enhanced to handle the various relations of influences.