https://ir.unisa.ac.za/handle/10500/460 Tue, 04 Aug 2026 08:20:57 GMT 2026-08-04T08:20:57Z https://ir.unisa.ac.za/handle/10500/32818 I am a “Thou” Therefore “I am”: Embedding Values and Norms into Intelligent Systems Nkohla-Ramunenyiwa, Thando; Mnkandla , Ernest The Enlightenment Era ushered in a platform for science to mark its epistemological territory, which has gained a following in a number of disciplines. One of these disciplines is technology. Technological advancements have progressed exponentially, to the point that on the ethics front, it has become challenging to catch up with this growth. Intelligent systems are a dimension of technological advancements that have, on the one hand, brought a considerable amount of convenience for society and, on the other hand, presented ethical concerns that require attention. At times complementary with capitalistic ideals, intelligent systems seem to be more appealing to the consumer than to the individual as a moral agent. Consequently, this paper aims to recommend a way forward when it comes to embedding values and norms into intelligent systems. Mon, 29 Jul 2019 00:00:00 GMT https://ir.unisa.ac.za/handle/10500/32818 2019-07-29T00:00:00Z https://ir.unisa.ac.za/handle/10500/32778 Multi-lingual emotion detection in speech using hybrid neural network architectures Norval, Michael John The advancement of Deep Learning has revolutionised numerous fields, including humancomputer interaction (HCI) and emotion detection. In speech recognition, voice assistants such as Apple's Siri, Microsoft's Cortana, and Google's Assistant have become integral to everyday technology use. However, current emotion detection systems often need more support for African languages, presenting a significant research gap. The aim and objectives of the study are to develop a Speech Emotion Recognition (SER) system tailored to South African languages, with a specific focus on Afrikaans. The primary objective is to compile an Afrikaans speech corpus for emotion recognition. Furthermore, it explores hybrid neural network architectures combining CNN, Recurrent Neural Networks (RNNs) and Long Short-Term Memory (LSTM) networks for improved accuracy. Finally, the aim is to investigate the optimisation of SER models for multilingual cross-language support. This study adopts a pragmatic research paradigm, utilising the Design Science Research Framework (DSR). Data is collected quantitatively. The literature review covers human speech, speech physiology, neural network architectures, speech corpus, and data extraction methods. Hybrid architectures and preprocessing techniques are examined to identify the most effective configurations. The Afrikaans speech corpus is sourced from Creative Commons (CC) sources, with data augmentation using Generative Adversarial Networks (GANs). The study successfully compiled an Afrikaans speech corpus. Synthetic speech samples were created using frameworks such as Tacotron 2 and the WaveNet vocoder, thereby enhancing sound quality. For the Hybrid Neural Network Architectures: The novel Dendritic Convolutional Long Short-Term Memory (DCLSTM) architecture outperformed traditional CNN and LSTM models, effectively capturing nuanced and long-term emotional dependencies in audio data. The second Kalman filter variant was identified as the optimal preprocessing technique for noisy speech signals. A final accuracy of 78.50% was achieved, improving on baseline CNN and LSTM models. For the Multilingual SER System, the DenCaps model achieved up to 71.91% accuracy on multilingual datasets, demonstrating improved generalisation and better capture of temporal dynamics and spatial hierarchies in emotional speech. The findings highlight the effectiveness of hybrid neural network architectures and advanced preprocessing techniques in improving SER accuracy. The development of the Afrikaans speech corpus and the introduction of the DenCaps model represent significant advancements in emotion detection. These results underscore the potential for creating robust, multilingual SER systems that cater to diverse linguistic needs. The research objectives were successfully achieved through the development of a specialised Afrikaans speech corpus, the exploration of innovative neural network architectures, and the optimisation of models for multilingual support. The academic contributions pave the way for more accurate, contextually aware human-computer interaction systems that address the research gap in emotion detection for African languages. Text and abstract in English Wed, 01 Jul 2026 00:00:00 GMT https://ir.unisa.ac.za/handle/10500/32778 2026-07-01T00:00:00Z https://ir.unisa.ac.za/handle/10500/32777 A quantified self technology framework for monitoring and managing diabetes Mutunhu, Belinda Background: Diabetes is a chronic disease that requires constant monitoring of blood glucose levels. However, in developing nations, most patients with diabetes fail to manage this disease. This is chiefly because the prevailing traditional methods involve infrequent hospital visits and the use of glucometers, spreadsheets, and logbooks. This is a tedious and inaccurate process that lacks real-time decision-making. Thus, most patients fail to be actively involved in the management of the disease, leading to serious complications. Quantified self-technology (QST), which uses wearable devices and mobile applications to monitor physiological parameters, offers an interactive, integrated solution for managing diabetes. However, owing to a lack of expertise, sociocultural challenges, and insufficient tailored theoretical frameworks, the adoption of QST is minimal in developing nations. Objective: The main objective of this study was to propose a framework for adopting quantified self-technology for managing diabetes in developing countries. Methodology: To understand QST, a systematic literature review identified the components of the QST framework. Resultantly, a conceptual framework was developed, which served as a guide in designing the research instruments for data collection. Thus, this study employed a single, embedded case study design and face-to-face interviews for data gathering. An interpretivist paradigm was used to interview 35 patients with diabetes from a hospital in Bulawayo, who were purposively and self-selected. Braun and Clarke's thematic analysis, facilitated by ATLAS.ti software, was used to improve methodological rigour. Results: The six core drivers and inhibitors of QST adoption among diabetics were identified as user characteristics, technology preparedness, perceived benefits, usability, social norms, and cybersecurity risks. Key user characteristics include digital literacy, self-efficacy, and awareness. Technological preparedness depends on cost, access to resources, training, and lowering technology anxieties. Over 70% of participants had a tertiary education, yet 69% were unaware of QST before the study. Participants perceived usability to be determined by usefulness, ease of use, and enjoyment. Furthermore, social norms can encourage or discourage adoption, depending on the views of the family and social surroundings. Cybersecurity risks may include privacy breaches, financial losses, and the misuse of information. Conclusion: This study proposes a framework for adopting QST to enhance diabetes self-care in developing countries. With the help of QST, patients can prevent acute complications, reduce hospital admissions, and be proactive in managing diabetes. The cybersecurity risks identified must be addressed to build trust and unlock QST benefits, including improved monitoring, adherence, motivation, data sharing, and cost savings. Thus, national health strategies should integrate QST into preventive care policies, as it will aid in monitoring, patient empowerment, and personalised treatment. However, the findings should be interpreted within the context of the study’s limitations, including self-selection bias, the single-hospital setting, and the focus on one QST application. Contributions: This research contributes to the field of health informatics theoretically. This is achieved through, first, offering a new conceptual framework for the adoption of QST within a developing country context and, secondly, identifying two new determinants, namely training and technology anxiety, and demonstrating the contextual operationalisation of digital literacy and motivation in a developing-country diabetes-management context. It also highlights the importance of explicitly recognising perceived benefits in adoption theory, demonstrating that understanding value is essential before introducing new technologies among patients. Practically, the framework provides actionable guidance for designing low-cost, user-friendly QST interventions suited to resource-constrained settings. Lastly, the framework is particularly valuable to healthcare providers, developers, and policymakers who seek to improve diabetes management through innovative self-monitoring solutions. Text and abstract in English Sun, 01 Feb 2026 00:00:00 GMT https://ir.unisa.ac.za/handle/10500/32777 2026-02-01T00:00:00Z https://ir.unisa.ac.za/handle/10500/32696 Ai-tutor framework for online programming courses Van den Berg, Yunduan This research report presents the design, development and evaluation of a proposed AI-Tutor framework designed to augment the capabilities of online tutors in supporting learners in introductory programming courses. The proposed framework addresses challenges such as limited personalised feedback, debugging difficulties, and reduced learner engagement. The study used a design science research methodology to systematically develop and evaluate the proposed AI-Tutor framework. A theoretical framework was developed to guide the design and development of the proposed AI-Tutor framework. A systematic literature review was conducted to identify the proposed framework's design criteria, components and requirements. The proposed AI-Tutor framework incorporates features, including personalised feedback, automated debugging tools, and interactive learning activities. These features are designed to provide a more engaging and effective learning experience for learners enrolled in online IPCs. A panel of experts evaluated the framework’s usability, practicality, and potential impact, providing qualitative feedback that informed refinements. Analysis revealed the potential to improve learner outcomes, and enhance understanding, problem-solving skills, and satisfaction. This research contributes to the online programming education by demonstrating the AI’s potential to aid the learning experience and offering a practical framework for educators, developers and researchers seeking to enhance the online learning experience. Sat, 01 Mar 2025 00:00:00 GMT https://ir.unisa.ac.za/handle/10500/32696 2025-03-01T00:00:00Z