On The Control Of Opinion In Social Networks

On Behaviour-Based Network Inference and Distribution-Based Network Control

“…A significant portion of the existing network control research, has focused on moving a system between arbitrary state vectors (Liu et al, 2011b;Ruths and Ruths, 2014). The NCP work of Runka (2016) expanded upon this vector-to-vector formulation by defining a control utility function based on an aggregate measure of the system state. This thesis moves beyond a vector-space model, proposing a distribution-based control solution (see Section 5.2), in which the target for control is no longer a vector, but is represented by the distribution of state values.…”

“…however, this is not the only possible view of network control. More recent work within the network control domain (e.g., Runka, 2016;Runka and White, 2015a;McKenney and White, 2016) has considered the behavioural aspects present within many networked systems, directing the focus toward a more practical view of network control. This opens up the possibility of many different types of network control problems involving failure avoidance, the limitation of state change, or other control goals.…”

“…This opens up the possibility of many different types of network control problems involving failure avoidance, the limitation of state change, or other control goals. By moving beyond the full state controllability view of network control, new targets for control and state representations can also be realized, such as the aggregate measures used to determine failure in the work of Runka (2016). These types of representations and control 1 Within this work, the terms 'network' and 'complex network' are used interchangeably.…”

“…The existing Network Control Problem (NCP) work has investigated a number of different control node set selection algorithms (Runka and White, 2015a;Runka, 2016), but there is still a need for more in depth analysis to determine the properties of an optimal control node set and identify methods for generating these sets. An additional problem that must be addressed within network control research is the selection of signals to inject into the system, or the selection of other control methods, such as the limitation of information flow, in order to achieve the desired control goals.…”

“…The work of Runka (2016) uses neural networks successfully for this purpose, but there still exists a significant area of work that should aim to understand the effect of control signals within networked systems and develop improved control signal solutions.…”