Theses and Dissertations (Informatics)

Permanent URI for this collectionhttp://hdl.handle.net/2263/31927

Browse

Now showing 1 - 20 of 177

Machine learning for social event analysis : public perceptions during COVID-19 and collective violence in South Africa
Kekere, Temitope (University of Pretoria, 2025-09-30)
Social media data is a rich source for understanding social phenomena. The computational analysis of the dataset provides a different perspective on the observed phenomena. The data provided on social media platforms allows users to communicate and engage in discourse without the restrictions that traditional surveys and polls sometimes impose. Although surveys and polls are not always restrictive, social media data provides an opportunity for freer public discourse on any social event. The advancement in natural language processing, computing power, and pretrained language models has given rise to advanced text analysis. Computational analysis of social media complements the sociology framework that theorises human behaviour during a public health crisis or societal instability. Using a mixed-methods research design, the study aimed to explore how machine learning grounded in social theories can enhance the understanding of human compliance, adjustment and collective violence. The study analysed Twitter conversations (currently known as X), combining natural language processing from computer science with social theories from sociology to explore, explain, and interpret human behaviour during the COVID-19 pandemic and the July 2021 unrest in South Africa. To be specific, sentiment analysis served as a proxy for understanding public perception and compliance with non-pharmaceutical interventions by the South African government during the pandemic in the first study. In the second longitudinal study, sentiment analysis provided an empirical evaluation of adjustment phases during the pandemic. The third study showed the utility of sentiment analysis in measuring the diffusion of collective behaviour during the jailing of former President Zuma, which sparked the unrest. In addition, topic modelling enabled the in-depth exploration of discourse that occurred during compliance with government policy during the COVID-19 pandemic, the adjustment phases, and the spread of unrest in South Africa. The study produced two gold-standard datasets to support the findings. Humans created one dataset, while a pre-trained language model generated the other. The production process is outlined to demonstrate reproducibility in other scenarios. The datasets are a valuable resource for computational scientists and sociologists advancing the study of human behaviour in similar or different social events and contexts. The study's introduction of sentiment analysis as a social marker of compliance, adjustment and collective behaviour highlights empirical evaluation of sociological theories and how computational analysis contributes to the growing field of sociology. The findings of the first study showed widespread negative sentiment, indicating a lack of public confidence in the South African government’s response to the pandemic. The second study showed that the adjustment phases to non-pharmaceutical interventions during COVID-19 were complete and followed the W-curve adjustment model, although the curve was inverted. By tracking sentiment over time, the study introduced sentiment as a measure of human adjustment and suggested that the interval between the four adjustment phases seems to be within 3 months. The third study showed that the spread of violence during the July unrest followed an S-curve diffusion pattern in sentiment. These findings underscore that social media serves as an early indicator of public health and urban safety crises, and computational analysis complements sociological frameworks. The machine learning models provided in the study, when interpreted within sociological frameworks, can provide monitoring of compliance, adjustment, and diffusion. This integral approach provides feedback on government policies and policing, serving as an evidence-based strategy for maintaining safer cities and healthier societies.
A framework to understand the role of generative AI to enhance customer experience and personalisation in marketing
Coetzer, Willem Adriaan (University of Pretoria, 2026)
Organisations are trying to utilise Generative Artificial Intelligence (AI) to improve customer experience and personalisation in marketing. However, the lack of clear guidance results in the production of unavoidable risks and inconsistent outcomes. What is missing is a coherent, evidence-based framework that guides the understanding of the role of generative AI to enhance customer experience and personalisation in marketing. Literature emphasises that personalised, trust-centred stakeholder interactions across service ecosystems result from customer experience advantages. In contrast, the unguided use of generative AI could damage brand trust and fail to deliver any meaningful improvement. This study applied a qualitative research design using semi-structured interviews with 15 marketing professionals The data was analysed though a deductive thematic analysis guided by the Service Dominant Logic (SDL) theory. Eight key themes emerged from the data and were synthesised into a seven-layered framework framework steered by the SDL theoretical lens that clarifies the organisational preconditions, decision-making gates, AI-enabled capabilities, customer touchpoints, experience outcomes, governance controls, and learning feedback loops necessary for value co-creation. The findings reveal that generative AI acts as a process enhancer rather than a stand-alone tool, creating value when foundational enablers are in place and when generative AI implementation initiatives are guided by structured decision gates. The framework operationalises the SDL constructs of value co-creation, service ecosystems, and marketing, within an AI-mediated marketing context, and demonstrates how these concepts can be applied in practice through ethical governance and continuous learning. Practically, the framework provides marketers with a systematic model for responsible generative AI implementation, verifying enablers, applying decision gates, configuring capabilities, aligning touchpoints, measuring value-in-use, and scaling only when outcomes are stable. This research advances understanding by translating SDL principles into actionable guidance for AI-enabled marketing. It shows that generative AI enhances customer experience and personalisation only when it is strategically integrated, ethically governed, and continuously refines within a co-creative service system.
ChatGPT in business analysis : a human-AI collaboration perspective on augmentation and professional practice
Abbas, Rocky (University of Pretoria, 2025)
The growing presence of generative artificial intelligence tools such as ChatGPT is reshaping professional work practices, yet there is still limited empirical evidence on how they are used in business analysis. While organizations are rapidly adopting AI technologies, little is known about how business analysts (BAs) integrate ChatGPT into their daily activities, the value it offers, and the challenges it presents. This is critical, business analysis relies on human judgment, contextual understanding, and ethical responsibility. The rise of GenAI therefore raises questions about trust, accountability, and professional competence, making it essential to understand how BAs navigate its use in practice. To address this problem, the study adopted a qualitative research approach. Fourteen BAs, across sectors including finance, logistics, healthcare, mining, and government, were interviewed using semi-structured interviews. Data-analysis followed a two-phase process: first, a deductive thematic analysis guided by the research questions, and second, the application of the Human-AI Collaboration (HAC) lens to interpret the findings. Results show that BAs mainly use ChatGPT as a support tool for routine and low-risk tasks such as drafting documents, summarizing information, clarifying language, and generating initial ideas. Reported benefits include time savings, improved clarity, and reduced mental effort. However, key limitations involve accuracy, lack of real time data, weak contextual awareness, and limited domain knowledge. Trust in ChatGPT varied with task risk, and BAs consistently retained final control over outputs, especially in regulated or high stakes contexts. Ethical concerns around data privacy, accountability, and compliance also influenced usage. Overall, ChatGPT’s role remains within automation and augmentation, rather than collaboration. While it enhances efficiency and supports analytical work, it does not replace the core human dimensions of business analysis. Effective use of GenAI therefore depends on strong human oversight, ethical awareness, and adaptive judgment. Future BAs will need to combine analytical expertise with AI literacy and governance awareness to ensure that automation strengthens, rather than weakens, professional value.
Ethical considerations of society 5.0
Griffiths, Luke John (University of Pretoria, 2025-12-10)
Society 5.0 is the vision of a data-driven, human-centred society that uses advanced technology to address challenges such as sustainability and economic inequality. A key feature of Society 5.0 is the coming together of cyberspace and physical space through the integration of 4th Industrial Revolution technology into systems that affect all of society. The use of Industry 4.0 technology such as artificial intelligence and big data means that the ethical challenges present in those technologies currently may also be present in Society 5.0 implementations. There also exists the potential for emergent ethical issues not widely seen in the current industrial context of these technologies. This pragmatist, mixed method study focused on identifying and understanding the ethical considerations of Society 5.0. The study collected data using a survey strategy. A systematic literature review focusing on the ethical considerations of the advanced technologies required for Society 5.0 was conducted to inform the creation of a questionnaire. The questionnaire was self-administered online using Google Forms and was used to collect quantitative data. Semi-structured interviews were then conducted using purposive sampling of technology experts to collect qualitative data and provide context to the ethical considerations identified in the systematic literature review. The quantitative data was analysed through tallies, graphs and descriptive statistics. The qualitative data was transcribed from audio recordings using Microsoft Word and then analysed by using axial coding to identify themes for a thematic analysis. The result of this study was a categorised list of ethical considerations of Society 5.0, informed by the perceptions of technical and non-technical members of society. The expected outcome was that organisations will be able to use this categorised list as a reference frame for considering the ethics of future Society 5.0 endeavours. It was also expected that this study will serve as a frame for future research by providing an ethical perspective of Society 5.0.
A human-centred framework for the development and evaluation of online systems applying artificial intelligence
van den Berg, Cindy (University of Pretoria, 2026-11-15)
Artificial intelligence offers numerous benefits, but can also cause harm when not adopted through a human-centred approach and addressing concerns regarding artificial intelligence (AI). This approach is known as human-centred AI (HCAI) and aims to augment human capabilities to create safe, reliable, and trustworthy solutions. The primary goal of HCAI is to enable a significant degree of human control and automation, allowing humans to influence decisions and events. Although definitions exist for HCAI, a consensus is required on how HCAI solutions are designed and evaluated. An HCAI framework needs to be employed for the development of AI solutions to ensure an ethical solution, where the results are objective, credible, unbiased, and fair. The use of a framework will facilitate uniformity in strategies, principles, and goals in organisations. This study employed pragmatism to understand what factors constituted HCAI, the existing frameworks for HCAI, how organisations apply human-centred design, and how organisations can effectively implement a HCAI framework. The knowledge gained helped develop a framework that can be applied to achieve human-centred design for AI solutions through a DSR approach. The DSR cycles of awareness, suggestion, development, evaluation, and conclusion were followed, utilising systematic literature review, questionnaires, and semi-structured interviews as data collection methods. The data were analysed using thematic analysis and descriptive statistics, including mean, mode, median, and standard deviation. The resulting framework consists of two sections: development and evaluation. The comprehensive framework comprises 15 categories with humans as the focus and humancentredness as part of the organisational strategy. The organisational strategy includes accountability measures to strengthen the human-centredness of the organisation. Context and requirements gathering are required, incorporating HCAI consideration, before design and development are approached. The approach of design, data, model, and analysis is suggested with testing, prototyping, and communication as an iterative loop. Through the design, data, model, and analysis approach, the following categories of grouped components need to be considered: human-centred values and ethics, UX and human interaction, data and model governance, technical robustness and performance, and AI system capabilities and design. The operationalisation of the framework guides the design, development, and evaluation of AI solutions to ensure human-centred AI solutions that are reliable, safe, and trustworthy.
A framework towards the adoption of interactive technology within higher education
Coetzee, Dané (University of Pretoria, 2025-11-15)
Students within the higher education system were thrown a curveball when the demand for virtual teamwork increased significantly. These students were drastically forced to find innovative ways to complete their university work online. On the positive side, this allowed students to explore and familiarise themselves with the online world, which equipped them with a whole new set of skills in both technology and teamwork. Due to the increase in the incorporation of technology within higher education, it is important to understand how the integration and utilisation of technology enabled students to perform team related activities. The use of the communication and collaboration technologies within higher education can allow students to work in a hybrid team setting. When students work together in person, their relationships and personal connections will strengthen naturally. This helps the team work more effectively and ultimately reduce conflict and feelings of isolation. Students did not necessarily possess all the skills and knowledge to optimise virtual and hybrid teamwork. This research study acknowledges these challenges and uncertainties and provides a conceptual framework whereby students can analyse their team needs and adopt the correct technology to optimise their team’s needs. The conceptual framework is based on constructs from the Team Effectiveness Model (TEM), Unified Theory of Acceptance and Use of Technology (UTAUT), Technological Pedagogical Content Knowledge (TPACK) and Technology Acceptance Model Frameworks (TAM). The qualitative and quantitative data collected in this study was collected via an online questionnaire that was shared with third year and honours BCom Informatics students at the University of Pretoria. The study identified common obstacles faced by the third-year and Honours BCom Informatics students during virtual and hybrid team projects. Issues like isolation, communication hurdles, and managing conflicts stood out prominently in both the qualitative and quantitative data. These results highlight the importance of considering both social and technical aspects in collaborative work within technology-driven educational settings. They emphasise the necessity of aligning team tasks with the appropriate technologies to enhance collaboration. Addressing these challenges led to the creation of the Hybrid Teamwork and Technology Framework (HTTF), offering a useful strategy to enhance teamwork outcomes in hybrid learning environments. The outcome of this research study provides students within the higher education system to analyse and adopt a technology that is the best fit for their specific type of team and therefore their specific team needs. The discovery made throughout this research study is that the constructs of a successful hybrid team in higher education that integrated technology into their teamwork. It is, therefore, recommended that tertiary students take this framework and actively apply it to their circumstances the next time they have to choose a technology for their teamwork requirements. This study makes an applied contribution to the field of Informatics through the development of the Hybrid Teamwork and Technology Framework. Rather than building new theory, the study integrates existing concepts related to teamwork, technology use, and student experiences in virtual and hybrid learning environments. The framework is grounded in empirical findings from both qualitative and quantitative data and provides a practical decision-support lens for aligning teamwork needs with collaborative technologies in higher education.
The effects of chatbot usage by businesses on online customer service satisfaction in South Africa.
Boikanyo, Orebotse (University of Pretoria, 2025-10-10)
The fourth industrial revolution has introduced various technological innovations, with artificial intelligence (AI) at the core in influencing and transforming how South African businesses operate in customer service. There has been a rapid increase in the adoption of AI-driven tools, and chatbots have been no exception. This study examines the impact of chatbot usage by businesses on online customer service satisfaction in South Africa, providing insight into the relationship between chatbot adoption and customer satisfaction, grounded in the Technology Acceptance Model (TAM) as the theoretical framework, extending research into the South African customer service context outside the fintech, health and telecommunication industries, by exploring how the TAM constructs, such as perceived usefulness and ease of use of chatbots, influence user satisfaction. Data was collected from South African users who had interacted with chatbots within 12 months of the study. The key findings were 80% of users found chatbots user-friendly, 77% reported that chatbots were helpful with their customer service queries, and 58% experienced improved customer service through chatbots. In contrast, only 22% of users prefer chatbots over live human agents. Highlighting the preference for human agents for complex tasks and scenarios. The results highlight the opportunities and limitations of chatbot technologies in customer service. Although chatbots offer 24/7 access and efficiency, they fall short with complex customer queries. To address the gaps, the study proposes an 8-stage chatbot implementation framework covering alignment, capability assessment, user interface design, integration, handover models, deployment, transparency, and continuous improvement, guiding businesses in automation with customer expectations in South Africa’s evolving digital landscape.
Basic education and artificial intelligence : a South African framework
Ndhlovu, Bheki (University of Pretoria, 2025-09-30)
This study provides a conceptual framework for incorporating artificial intelligence (AI) into basic education in South Africa. There is a gap in AI implementation strategies and frameworks, particularly in South Africa. This study examines the advantages of AI, as well as barriers to acceptance in developing countries, utilising the Unified Theory of Acceptance and Use of Technology (UTAUT) model. Using the qualitative method, semi-structured interviews are conducted with educators in the basic education system to collect data. The results show that the basic education system in South Africa, particularly in the rural areas, does not have the basic infrastructure for technology use and is therefore not ready for AI. There are also obstacles in the form of funding and government support in terms of developing policies and budgets to help steer the implementation of AI. The implications of the data suggest that, for any implementation of the AI to be successful in South Africa, government needs to provide urgent support in the form of recurring budget allocations, which the framework links to educators’ behavioural intention to adopt AI, alongside policy interventions. This study recommends the collaboration of educators and the Department of Basic Education to create clear policies, invest in teacher training and encourage collaboration between all stakeholders, guided by a South African specific framework that goes beyond generic international models by specifying the school level enabling conditions required for equitable and sustained classroom use of AI.
A model for the application of the Scaled Agile Framework (SAFe) and its influence on project management practices in South Africa’s tier one financial institutions
Shai, Johannes Mashao (University of Pretoria, 2026-01-20)
Agile methodologies have transformed the global technology landscape by enhancing organisational efficiency, fostering collaboration, and improving product quality, signifying a significant shift in project management practices. Initially designed for small, co-located teams, Agile is now widely adopted by large organisations seeking to respond swiftly to dynamic customer needs, drive innovation, and maintain competitiveness. The Scaled Agile Framework (SAFe) is a leading approach for scaling Agile in complex, organisation-wide settings. However, South Africa’s tier-one financial institutions face unique challenges in applying SAFe, including fragmented adoption patterns and limited practical guidance. Furthermore, there is a lack of empirical research on the influence of SAFe on project management practices within the financial services industry. This study examines the factors influencing the application of SAFe and its influence on project management practices within South Africa’s tier-one financial institutions. Grounded in the unified theory of acceptance and use of technology (UTAUT) and the diffusion of innovations (DOI) theory, the study adopts an interpretivist paradigm and a qualitative methodology. Semi-structured interviews were conducted with 25 purposively selected participants across various SAFe-related roles. Data were thematically analysed using ATLAS.ti. Key findings indicate that SAFe is implemented as a strategic framework to drive organisation-wide transformation, enhance delivery efficiency, and align requirements with broader business objectives. However, several barriers hinder its effectiveness. These include misaligned performance metrics between business and IT, inconsistent interpretations of SAFe, a lack of standardised governance, and a persistent reliance on traditional reporting structures. These factors collectively undermine cross-functional collaboration and obstruct meaningful Agile transformation. This study proposed a SAFe application model, which was tested and validated through a focus group approach. The framework incorporates constructs that participants identified as appropriate and relevant for practically applying SAFe within tier-one financial institutions in South Africa. In doing so, the study contributes to the growing body of Agile scholarship in developing economies and offers practical, context-specific recommendations for improving SAFe implementation. It further underscores the importance of a coordinated transformation strategy that aligns structural and cultural realities with Agile principles to drive sustainable and meaningful change.
The use of critical thinking to identify fake news in developing countries : a case study in Nairobi County, Kenya
Masavah, Vincent Mwendwa (University of Pretoria, 2025-12-11)
The proliferation of online platforms such as Facebook, Google, and X (formerly Twitter) has facilitated the rapid spread of fake news in recent years, raising significant concerns across various sectors of society. In developing countries, it has incited political and xenophobic violence, impeded vaccine distribution during pandemics, and provoked religious conflicts resulting in fatalities. This phenomenon is attributed to the widespread availability and use of mobile technology, the Internet, and social media, which have exacerbated the prevalence and risk of fake news. Although numerous studies on fake news have been conducted, they predominantly focused on the Global North. This study aims to address the urgent need to mitigate the spread of fake news in developing countries. A potential approach to studying and recommending solutions to this problem is the application of critical thinking to identify fake news. Therefore, this study investigates how can critical thinking be employed to discern fake news in developing countries.A qualitative study was performed that utilised the Choice Framework as the underpinning theoretical framework. Semi-structured interviews with individuals and focus groups were conducted in Nairobi County, Kenya. Data were collected from 55 individual participants and 13 focus groups. Participants were selected by means of purposeful and snowball sampling strategies. The data collected were analysed thematically using Atlas.ti and unpacked in terms of the constructs of the Choice Framework. This study found that employing critical thinking to discern fake news decreases the probability of individuals being susceptible to it. However, development is a choice - only individuals who choose to apply critical thinking to identify fake news will be able to achieve the desired outcome. Critical thinking can assist them to preserve the integrity of their social structures and to lead the life they value. Ultimately, the benefits of development can in this case be better achieved through the application of critical thinking.The practical contribution in this study includes the recommendation that digital and information literacy should be taught at schools, the teaching and application of critical thinking should be encouraged, policies, programmes and laws that curb the spread of fake news should be introduced, and finally a fact-checking mechanism for information verification should be created. The theoretical contribution in this study is the application of the Kleine's Choice framework to understand the use of critical thinking in identifying fake news in the Global South.The study’s limitations include that it was conducted only in Nairobi County, Kenya and that the focus was restricted to the application of critical thinking to identify fake news. Future studies should focus on replicating the study in other regions. In addition, future studies should consider other mechanisms or avenues to identifying fake news.
The digital platform resource framework : a resource-based approach to business model transformation for competitiveness in the digital economy
Goneos-Malka, Amaleya (University of Pretoria, 2025-12-12)
Advances in digital technologies have fundamentally reshaped competitive landscapes across industries, enabling new forms of value creation and driving business transformation. The dominance of digital platform-based firms, such as Apple, Microsoft, Nvidia, Amazon, Alphabet, and Meta, demonstrates the power of digital technologies to establish sustainable competitive advantage. For traditional incumbent firms, organisational inertia often impedes adaption to environmental change, where legacy structures keep them bound to outdated models. Excluding digital innovation threatens the survival of firms established in the pre-digital era, while embracing digital transformation offers opportunities for reinvention and relevance.Although multiple approaches to digital transformation exist, platform business models represent a possible option. Incumbent firms already possess bundles of tangible and intangible resources; however, the challenge lies in reconfiguring these resources and developing new capabilities to compete effectively in the digital economy. The absence of a structured framework to guide this transformation reduces the efficiency and success of such initiatives.This study, therefore, investigates the question: “How can existing traditional businesses transform into digital businesses using platform-based business model strategies?” Guided by the Resource-Based View (RBV) of the firm, the objectives of the research were: (i) to identify the strategic resources associated with platform-based business models, and (ii) to develop an artefact, the Digital Platform Resource Framework (DPRF), to support traditional businesses in transitioning to competitive digital platforms.A design science research strategy was employed as the methodology for this thesis. A systematic literature review identified thirteen strategic resources critical to platform business models, which were operationalised through RBV within the Business Model Canvas. Multiple case study analysis confirmed that successful platforms integrate these resources synergistically to create, deliver, and capture value. Key findings reveal that platforms continuously reconfigure resources in response to environmental change, embed data analytics across ecosystems to enable personalised experiences and innovation, and leverage network effects as scalable growth drivers. Diverse monetisation strategies, such as subscriptions, freemium services, advertising, and cross-subsidisation, further enhance value capture.Building on these insights, the DPRF was developed as a three-stage process: (1) conducting a needs analysis of platform resources in the existing business model, (2) operationalising these resources through an RBV lens, and (3) mapping the reconfigured business model on the Business Model Canvas. The DPRF was validated through a proof-of-concept study, which demonstrated both its practical application and its ability to guide transformation in a real-world context.The study contributes theoretically by extending Osterwalder’s Business Model Canvas to incorporate platform-based resource configurations, and practically by providing managers with a structured instrument for digital transformation. Ultimately, the research advances understanding of how traditional firms can transition into platform-based digital businesses to sustain competitiveness in the evolving digital economy.
A knowledge visualisation framework for information systems (IS) audit in organisations
Nyamela, Andile Ayabonga (University of Pretoria, 2025-12-01)
The increasing complexity of information systems (IS) audits highlights the importance of effective knowledge transfer and collaboration between auditors and information technology (IT) auditors. Traditional audit methodologies often rely on text-based documentation, leading to knowledge silos and inefficiencies in decision-making processes. Knowledge visualisation (KV) offers a solution by integrating visual representations to improve communication, understanding, and retention of critical audit knowledge. This study aimed to examine the applicability of KV in IT audits by developing a KV framework tailored for IT audits, identifying key visualisation elements that improve audit execution and decision-making.Guided by an interpretivist research philosophy, this study used a mixed-methods approach to capture qualitative and quantitative perspectives on KV in IT audits. The research design used a questionnaire to understand current visualisation practices, challenges, and opportunities across different organisational sectors. A structured questionnaire was distributed via email to capture professional perspectives on KV in IT audits. The quantitative data enabled statistical analysis of trends and patterns, while qualitative responses provided deeper insights into auditors’ experiences, perceptions, and contextual factors that influence the adoption of visualisation. The study ensured a representative selection of participants through a structured sampling approach, thus ensuring the reliability and applicability of the findings within the IT audit domain. The study reveals that KV significantly improves knowledge transfer, collaboration, and audit decision making. Auditors using visual tools, such as process maps, conceptual diagrams, and interactive dashboards, report an improved understanding of complex audit findings and stakeholder communication. The study identified key challenges to KV adoption, including organisational resistance, a lack of training, and technological constraints. In addition, it highlighted best practices for integrating visualisation frameworks into IT audit processes, emphasising the need for tailored visualisation techniques aligned with audit objectives and organisational contexts. By providing a structured KV framework, the study offers practical recommendations for organisations aiming to improve their IT audit functions. Ultimately, this research contributes to the evolving discourse on KV by demonstrating its tangible benefits in auditing environments, paving the way for future studies to examine innovative visualisation methodologies that can further improve audit effectiveness and knowledge management.
A conceptual framework supporting user experience designers in creating conversational user interfaces
Rossouw, Amore (University of Pretoria, 2025-09-01)
Conversational user interfaces (CUIs) driven by Artificial Intelligence (AI) are becoming more common in a user’s day-to-day digital interactions. AI can be used to automate repetitive tasks, provide insights from large datasets, and enhance existing workflows. Furthermore, AI-driven CUIs such as AI chatbots and digital assistants can leverage technologies such as machine learning (ML) and natural language processing (NLP) to engage with customers through natural dialogues. Organisations can leverage user experience (UX) to ensure that the CUIs they utilise provide a seamless and satisfying experience to their customers, leading to a competitive advantage. Structured design principles and frameworks are needed to support the development of CUIs. Despite the growing interest in CUIs, UX designers continue to face challenges when designing CUIs due to the absence of context-specific CUI UX metrics and an oversight of the importance of integrating UX design practices with data science expertise. These gaps result in a mismatch between user expectations and the capabilities of CUIs, highlighting the need for comprehensive tools and guidelines to support UX designers in CUI design. The purpose of this study is to identify UX components and develop a conceptual framework that can assist UX designers in designing a CUI and communicating the design components of an effective CUI. Therefore, an interpretivist philosophical approach has been identified as the most suitable qualitative research paradigm. To capture both the subjective experience and the broader pattern across participants, this study adopted a mixed-methods approach. This study employed a design science research (DSR) research strategy, which outlines a structured approach for developing and evaluating artefacts. Technology-Organisation-Environment (TOE) was chosen as the theoretical framework. Based on the findings of this study, TOE was enriched with a fourth factor, namely Human Experience (TOEHE). In the context of CUI design, the findings in this study indicated that the role of a UX designer is to ensure that the CUI is engaging, intuitive and able to deliver a highly satisfactory UX. Furthermore, the role of a conceptual framework in categorising CUIs is defined as a communication bridge between different stakeholders involved in developing a CUI, ultimately leading to a more cohesive product. Additionally, the findings emphasise the importance of context-specific UX principles such as simplicity, personalisation and emotional considerations in designing engaging and intuitive CUIs. The final artefact, TOEHE, is presented after enriching the framework with two DSR sub-cycles. TOEHE comprises four main CUI adoption factors (technology, environment, organisation, human experience) and four unique themes (chatbot, user, effect, model) that are mapped to the CUI adoption factors and 28 individual concepts categorised under the four themes. TOEHE highlights the key components that influence the UX of CUIs. The framework presents a high-level, holistic overview that supports UX designers and developers throughout the design process.
Proactive insider threat management in Namibian state-owned enterprises
Shikonde, Selma Hambeleleni (University of Pretoria, 2025-12-09)
Over the years, the technological landscape has advanced, changing the entire security posture of organisations. This has led to organisations being exposed to security threats mostly caused by insiders. Insider threats pose various challenges to State Owned Enterprises (SOEs) globally, with Namibia facing vulnerabilities owing to limited cybersecurity infrastructure, expertise, and regulatory frameworks. The December 2024 Telecom Namibia breach, involving insider assisted data theft and ransomware, highlights the severity of this problem in Namibian SOEs. The existing literature has examined insider threats in private sector organisations, resulting in a gap on how SOE governance structures influence insider threat management in developing countries. There is also a gap in existing frameworks not addressing the SOEs constraints such as political oversight, resource limitations, and complex governance structures. Furthermore, existing insider threat detection approaches lack integration of behavioural analysis models with organisational maturity assessments tailored to environments with limited resources. The objective of this study is to propose a framework and strategies to proactively manage insider threats by implementing robust security and privacy measures. The study addresses five research questions namely, examining insider threat types, patterns and taxonomy, privacy impacts, current security measures, policy gaps, and proactive management strategies in SOEs. This study followed a Design Science Research (DSR) approach and developed the Insider Threat eXplainable Machine Learning (IT-XML) framework. This framework integrates the Cross-Industry Standard Process for Data Mining (CRISP-DM) methodology and Hidden Markov Models (HMM). The study utilised quantitative methodology using a structured online questionnaire. The survey data was derived from 60 participants across three Namibian SOEs. The IT-XML framework effectively addresses insider threat management in Namibian SOEs. It integrates CRISP-DM and HMM for proactive threat detection and organisational security maturity classification. The survey results revealed key insider threats, including information sharing violations (61.7%) and unauthorised access (46.7%). The framework achieved high classification accuracy (91.7%) and provided actionable security control priorities, such as audit log access limits and vendor breach notifications. The findings demonstrate that SOEs can adopt collaborative, evidence-based approaches to enhance cybersecurity. This study contributes scientifically through the creation of a dataset on insider threat management practices in Namibian SOE. In addition, utilising CRISP-DM and HMM to address insider threats in the context of SOEs is a scientific addition to the cybersecurity and data mining body of knowledge. The practical implications enable SOEs to implement collaborative security enhancement based on shared maturity profiles, moving away from individual organisational approaches towards coordinated capacity building. The framework provides specific security control priorities, enabling SOEs to improve their insider threat management capabilities with actionable, evidence-based measures that ensure better protection of critical organisational assets.
A framework for intra-organisational knowledge sharing towards business performance improvement in the retail banking industry in South Africa
Nkuna, Magogodi (University of Pretoria, 2025-11-02)
In the current economic climate, effective knowledge management is vital for organisational growth and development. The highly competitive South African banking industry compels banks to leverage existing knowledge to create unique, inimitable knowledge assets. Managing knowledge is essential because banks are viewed as not only selling services but also knowledge. The South African banking sector faces competition from new entrants, including FinTech organisations, that can innovate more easily due to their agility. Studies have shown that South African retail banks lack formal knowledge-sharingpractices, leading to a risk of losing valuable knowledge when employees leave. Organisations need to formalise this practice to utilise knowledge assets fully and benefit from knowledge sharing (KS). The purpose of the research was to develop an intra-organisational knowledge-sharing framework (IOKSF) for the South African banking industry. The study utilised a design science research (DSR) methodology with three distinct cycles to design the IOKSF. The first cycle involved comprehensive literature reviews to create the initial version of the framework. The second cycle included a survey with managers in retail banks, leading to the second version of the framework. The final cycle utilised focus groups to refine and produce the third version of the framework. Knowledge management (KM) experts evaluated the final version of the IOKSF, and the study incorporated their feedback and input to complete the framework. However, the framework was not piloted in a real world organisational setting, and its practical application remains to be tested. The IOKSF offers retail banks and other organisations a structured approach to implementing KS within their operations. This comprehensive framework details strategic and operational aspects, identifies potential barriers, and provides actions to mitigate these challenges. In addition, the study includes an implementation plan to guide organisations in adopting the framework effectively.
A theory of enterprise architecture management performance
Mbuya, Donald Penn (University of Pretoria, 2024-11)
Problem and objective: Enterprise Architecture Management (EAM) as a strategic management discipline has the potential to improve organisational performance. There are numerous claims that EAM benefits have a positive impact on organisational performance, yet empirical evidence is scarce and conjectural. Research articulating how EAM performance constructs interact synergistically to improve EAM performance itself and, hence organisational performance is elusive. In other words, the EAM value adding mechanism and dynamics to organisations is not only well understood but also insufficiently researched. The main objective of this research was to elucidate how EAM performance constructs interact harmoniously to improve the performance of EAM itself and thus, organisational performance consolidated into a holistic, analytical, explanatory, and predictive model or theory of EAM performance. Methodology: Relevant papers were identified via a systematic literature review and thematically analysed. Snowballing and purposive sampling techniques were used to collect data. Performance constructs from the task-technology fit theory, fit-viability theory, and organisational performance motivation theories were integrated to derive a parsimonious, yet comprehensive model of constructs that influence EAM performance and thus, that of an organisation. A global sample (N=243) was analysed using structural equation modelling to validate the model or theory. Research findings: Model fit results were as follows: RMSEA=0.04, SRMR=0.06, Chisq/df=1.29, CFI=0.90 and TLI=0.90) were within the acceptable range of model fitness. The results indicate that EAM job-fit EAM technology (p=0.048) and EAM organisational viability (p=0.000) are significant predictors of EAM performance. They also show that EAM performance (p=0.017) is a significant predictor of organisational performance. The correlation between EAM personnel-fit organisation (p=0.104) and organisational performance was inconclusive. These findings contribute to improving understanding of how EAM adds value to organisations thereby addressing a notable gap in current empirical research. Conclusion : EAM performance is a significant predictor of organisational performance. The study advances the epistemology of EAM by presenting a novel and significant contribution to the current understanding of EAM and its impact on critical organisational performance outcomes. Researchers are invited to validate and extend with it with a larger sample size (N≥300) as an established contemporary Information systems theory. Implications for practice and academia: The theory should be applied to measure and improve the performance of the EAM practice and thus, organisational performance. For academia, the predictive properties of the theory should be tested globally or in a unique context like an organisation or a country in the form of a case study. Modifying and extending the theory remains an opportunity for future research.Originality/Novelty: The theory is a creative and innovative integration of the task-technology fit theory, fit-viability theory, and organisational performance motivation theories applied in the context of EAM performance correlated with organisational performance.
The impact of culture on communication during user requirements gathering : a South African banking case study
Nyalungu, Dorcas Hope (University of Pretoria, 2024-11)
There are limited studies on how societal culture impacts communication during the user requirements gathering process. In the South African context, this issue is more prevalent than in other societies because of different dominant ethnic groups. The study explores the impact of societal culture on communication in the user requirements gathering phase of typical projects in the banking industry. The study sought to determine the role that different societal cultures in South Africa play in the communication processes involved in gathering user requirements within the banking sector. The study employed an interpretivism research philosophy and an inductive research approach. The research strategy was a case study strategy, which utilised a qualitative research method with semi-structured interviews. Sixteen business analysts (BAs) were interviewed. Thematic data analysis was employed to generate understanding and insights into the phenomenon under investigation. Several insights were identified during this study, key of these include - the importance of cultural sensitivity; the role of digital communication tools; and the need for inclusive communication strategies. The study participants emphasised the necessity of understanding and adapting to diverse communication styles, balancing directness with respect for hierarchy, and using culturally appropriate metaphors and idioms. The findings highlight the challenges posed by linguistic diversity and the significance of promoting cultural awareness and sensitivity training. There was also emphasis on the importance of considering different factors, such as communication style and cultural values, when understanding how users from various cultures share their needs. The key communication techniques for effective user requirements gathering were found to be pre-session preparation, openness, adaptation to stakeholders, emotional control, trust building, engagement maintenance, and goal setting. Recommendations include improving language proficiency, fostering cultural sensitivity, and investing in relationship-building efforts. Future research should explore broader geographic representation and use quantitative research methods to gain more understanding of the phenomenon under investigation.
Improving knowledge acquisition with obsolescence planning : the case of the Zambian civil service
Lukwesa, Katongo (University of Pretoria, 2024-09)
The study examines how the knowledge state affects performance in Zambia's civil service and proposes a solution through obsolescence planning. The research develops a framework that informs a model, supported by a structured template for data collection and analysis. The study's key contribution to academia is its introduction of new methods for addressing knowledge acquisition. In IT, it offers innovative techniques and models for managing knowledge obsolescence. For practitioners, it provides a model to improve knowledge acquisition strategies, ultimately enhancing decision-making and service delivery in the civil service.
Understanding the use of pejorative language in massively multiplayer online games (MMOGs)
Namane, Kedibone Charlotte (University of Pretoria, 2024-09)
Massively multiplayer online games (MMOG) contribute approximately 26 billion USD with approximately 1.1 billion players. Research has shown that many of the players are teenagers and young adults. Furthermore, these games can have a powerful influence on the players and their lives outside the online game. Ample research exists on the prevalence of toxic behaviour during gameplay, including bullying, harassment, and cheating. Less research is available on language use, and more specifically, the use of pejorative language during the playing of MMOG. Existing research on this topic indicates a gap in understanding the linguistic habits of gamers during gameplay. Exposure to toxicity and profanity (such as pejorative language use) during MMOG play can have repercussions such as relational aggression in adolescents. Therefore, the purpose of this study is to gain an understanding of MMOG players’ perceptions of pejorative language use in MMOGs. In addition, the study aims to understand how other players account for the strong presence of pejorative language and expletives in player discourse. The research study adopted a grounded theory methodology to develop a theoretical model, which draws on the stressor-strain-outcome (SSO) model, to understand the use of pejorative language in MMOGs by explaining different stressors that result in the use of pejorative language in MMOGs. This research took a qualitative approach in examining the main concerns of using pejorative language in MMOGs as perceived by the players. Twelve interviews were conducted before theoretical saturation was reached. These participants supplied rich data for the researcher to draw on due to their gaming experience. The data gathered led to the emergence of a substantive theory, “Understanding the Use of Pejorative Language in MMOGs”. The theory proposes that game language and competition (both inter-team and intra-team) are influenced by the type of game played. The influence of these core categories will, in turn, play a role in the actual language behaviour, which, in the context of this study, has been identified as pejorative. The actual language behaviour depends on the different gaming platforms and the moderation and medium of communication used in these platforms. Some outcomes then result from the actual language behaviour. The SSO model was used to explain and expand on the relationships between the core categories. This study's primary contribution to the body of knowledge is a theoretical understanding of MMOG players' perceptions of pejorative language use in MMOGs. To the best of the researcher’s knowledge, this study is the first to describe the use of pejorative language in MMOGs supported by qualitative data.
Consumer artificial intelligence impact on organisations' data analytics and business intelligence processes
Bownass, Britney (University of Pretoria, 2024-09)
This study examines people's intention to use consumer AI in data analysis and business intelligence. Introducing consumer AI in data analysis can make individuals more productive and efficient. However, some individuals fear how this innovation will change their employability. This study aims to determine which data analysis and business intelligence tasks consumer AI is best suited for, which characteristics of consumer AI make it appropriate for use in these tasks, and the factors organisations need to consider when introducing consumer AI in their data analysis and business intelligence processes. In meeting these research objectives, the study ultimately aims to determine to what extent consumer AI can be used in data analysis and business intelligence. This study adopted an interpretivist philosophy to understand the nuances. Qualitative data was collected through semi-structured interviews with fifteen analysts. A theoretical foundation was constructed by integrating three prominent theories, namely, the Unified Theory of Acceptance and Use of Technology (UTAUT), the Innovation Resistance Theory (IRT), and the Technology Organisation and Environment (TOE) framework. This theoretical foundation was used to develop the interview guide. The results showed that most participants believe consumer AI is best suited for data analysis. However, multiple participants indicated that consumer AI is useful in pre-processing data and visualising findings, ultimately increasing business intelligence and leading to better-informed organisational decisions. The participants identified eight characteristics that make consumer AI appropriate for data analysis and business intelligence. One of the main characteristics is that the chatbot is easy to use and that users can communicate with the application in natural language. The data revealed seven consumer AI drivers and six barriers ultimately impacting an organisation's adoption of consumer AI in data analysis and business intelligence.

Browse

Recent Submissions