Thesis
A study on the effects of a selfish agent in networked opinion dynamics
Washington State University
Master of Science (MS), Washington State University
2019
Handle:
https://hdl.handle.net/2376/103316
Abstract
This thesis is concerned with manipulation of a network opinion dynamics by a selfish agent. To study manipulation, a canonical model for network opinion dynamics with local linear averaging updates is extended to include a manipulator (selfish) agent. The selfish agent uses a projection (statistic) of network opinions in feedback to regulate the dynamics, with the goal of driving other agents' opinions to a desired reference or goal. Specifically, the selfish agent in this work is modeled as either a proportional or proportional-integral feedback controller. An extension of the model with a nonlinear update rule, wherein all agents' opinions are capped or held between 0 and 1, is also considered. Analytical results are provided characterizing consensus behavior of the manipulated opinion dynamics model, including in the presence of a second stubborn agent with a fixed opinion. Additionally, an empirical study is conducted to identify the most influential nodes, and a novel centrality metric is proposed that closely correlates to the influence metric. Finally, simulations if the manipulated opinion dynamics are developed for a variety of scenarios. The empirical analyses and simulations were conducted using a real-world social network, representing frequent interactions between dolphins in the Doubtful Sound, a remote Fjord in New Zealand.
Metrics
6 File views/ downloads
19 Record Views
Details
- Title
- A study on the effects of a selfish agent in networked opinion dynamics
- Creators
- Nathan Wendt
- Contributors
- Sandip Roy (Degree Supervisor)
- Awarding Institution
- Washington State University
- Academic Unit
- Electrical Engineering and Computer Science, School of
- Theses and Dissertations
- Master of Science (MS), Washington State University
- Publisher
- Washington State University; [Pullman, Washington] :
- Identifiers
- 99900525397101842
- Language
- English
- Resource Type
- Thesis