Improving the security and cyber security of companies and individuals using behavioural sciences: a data-centric approach

Castro Gonzalez, Leonardo Mariano ORCID: 0000-0001-5631-0137 (2023). Improving the security and cyber security of companies and individuals using behavioural sciences: a data-centric approach. University of Birmingham. Ph.D.

Preview

CastroGonzalez2023PhD.pdf
Text - Accepted Version
Available under License Creative Commons Attribution Non-commercial.
Download (7MB) | Preview

Abstract

While security and cyber security systems literature focus on how to detect threats at a logistics, software and hardware level, there is not enough work around how to improve the security by incorporating the understanding of the human behaviour for those individuals that form part of the system. The present dissertation focus in the latter problem and has it as main research question. To do so, we study three different security and cyber security problems. We study a problem of communication framing when training employees in cyber security by deploying a two-staged survey in a British financial institution to then analyse it with a behavioural segmentation model. We find that, depending on their risk-perception and risk-taking attitudes, employees can become better cyber security sensors when correctly framed. We also study a problem of illicit drugs distribution in England to understand the territorial logic of the operators. Using public data, we analyse the problem using Spatial Analysis models. We find that gangs avoid places with a high number of knife crime events and hospital admissions by misuse of drugs. Finally, we study the transition of companies to the “New Normal” when the pandemic started. Using a qualitative model to understand the cyber security culture within, we find that cyber security was not a priority of the narrative of big companies during the first months of 2020. The three essays contribute to the literature in behavioural sciences applied to security and cyber security by using modern tools and frameworks of statistical learning and Natural Language Processing. By incorporating these different resources, we show how to improve the efficiency of security and cyber security systems by analysing the behaviour data extracted from them.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):

Supervisor(s)	Email	ORCID
Pogrebna, Ganna	UNSPECIFIED	orcid.org/0000-0002-5487-7284
Scharf, Kimberley	UNSPECIFIED	UNSPECIFIED
Talavera, Oleksandr	UNSPECIFIED	orcid.org/0000-0002-4799-778X