Mosquera, Guillem (2020) Network science for social and technological systems. PhD thesis, University of Warwick.

Preview

PDF WRAP_Theses_Mosquera-Donate_2020.pdf - Submitted Version - Requires a PDF viewer. Download (38Mb) | Preview

Request Changes to record.

Abstract

This thesis contains a collection of research outcomes from the field of complex networks. The results presented here have been divided in two parts, one devoted to theoretical methods and the other to data-driven applications. Although many of the results, especially in the first part, are general enough for describing many complex systems, a special focus on social systems has been used throughout the thesis.

The first part contains ideas that explore the interplay of topology and dynamics in complex systems, divided in three chapters dedicated to opinion dynamics, modular networks and weighted networks respectively. Regarding opinion dynamics, we study the emergence of self-organised leadership and herding behaviour in the voter model. Regarding modular networks, we present a generative model for networks with community structure and arbitrary bridgeness distribution. We also show how bridgeness interplays with functional behaviour in different dynamical systems. We use such interplay to define the concept of dynamical centrality, and show its applications to network dismantling under limited topological information. Finally, we demonstrate how topological uncertainty in link weights induces fluctuations on the critical threshold for multiple dynamical processes on networks. We also discuss the role of degree heterogeneity in this propagation, finding non-trivial dependencies for scale-free networks.

The second part contains two applications of network analysis to real-world systems. The first application is a data study on the rail network of London and its surrounding area. We show how topological resilience measures are strongly correlated to the performance of train operators in the network. The second application contains a network-based model of armed conflict prediction at city level of analysis. We use several centrality measures as features for machine learning models, showing how network information generates very significant improvements in out-of-sample prediction performance.

Item Type:	Thesis (PhD)
Subjects:	Q Science > QA Mathematics
Library of Congress Subject Headings (LCSH):	System analysis, Dynamics, Topology
Official Date:	September 2020
Dates:	Date Event September 2020 UNSPECIFIED
Institution:	University of Warwick
Theses Department:	Centre for Complexity Science
Thesis Type:	PhD
Publication Status:	Unpublished
Supervisor(s)/Advisor:	Johnson, Samuel (Professor)
Format of File:	pdf
Extent:	xv, 141 leaves : illustrations (chiefly colour), maps (chiefly colour)
Language:	eng

Request changes or add full text files to a record

Repository staff actions (login required)

View Item

Downloads