Saudi Cultural Missions Theses & Dissertations
Permanent URI for this communityhttps://drepo.sdl.edu.sa/handle/20.500.14154/10
Browse
3 results
Search Results
Item Restricted Deepfake Technology: A Multi-Disciplinary Analysis of Technical, Social, Legal, and Ethical Issues(Saudi Digital Library, 2026) Alanazi, Sami; Asif, SeemalAdvances in deep learning have enabled the creation of highly realistic synthetic media capable of manipulating faces, voices, and actions with increasing fidelity. These techniques support legitimate uses in entertainment, education, and digital content production. At the same time, misuse has raised concerns linked to misinformation, privacy infringement, identity fraud, and ethical accountability. This thesis examines deepfake technology as a socio-technical problem, integrating technical analysis with social, legal, and ethical perspectives to inform effective mitigation. Three strands structure the investigation. A critical review of the literature establishes the state of generative and forensic research. Fourteen expert interviews across artificial intelligence, law, ethics, and cybersecurity provide institutional and professional context. Alongside this, empirical experimentation evaluates deep learning approaches to deepfake detection under realistic conditions. Initial technical work focused on image-based forensics, leading to the development of AI-Guard, a convolutional detection system designed for practical deployment. Evaluation results indicate that AI-Guard achieved 98% validation accuracy and 93.2% test accuracy on a dataset exceeding 450,000 images, while supporting real-time inference on mobile platforms. Video forensics forms the second technical contribution. Building on insights from earlier experiments, the thesis introduces VIDS-Guard, a forensics-aware detection framework that combines frequency-domain features, YCbCr colour decomposition, and temporal Transformer-based attention. Trained on 26,975 videos, the system achieved an accuracy of 0.91, a Macro-F1 score of 0.90, and an AUC of 0.97, outperforming established benchmark models across multiple evaluation settings. The results suggest that architectural diversity and temporal modelling improve generalisation under dataset variation. Technical findings do not stand alone. Analysis of expert perspectives and regulatory gaps indicates that detection systems, if unreliable or poorly governed, may themselves contribute to harm. Effective responses therefore appear to require coordination between deployable forensic tools, legal clarity, and ethical safeguards. By aligning system design with institutional and societal considerations, this thesis advances both practical detection capability and regulatory understanding, contributing to transparency, accountability, and resilience within the digital media ecosystem.10 0Item Restricted AN EXPLORATION OF EMERGENCY STAFF PERCEPTIONS AND EXPERIENCES OF TEAMWORK IN AN EMERGENCY DEPARTMENT IN THE KINGDOM OF SAUDI ARABIA(Cardiff University, 2024) Alanazi, Sami; Whitcombe, SteveBackground: Teamwork practices have been recognised as a significant strategy to improve patient safety, quality of care, and staff and patient satisfaction in healthcare settings, particularly within the emergency department (ED). The ED depends heavily on teams of interdisciplinary healthcare staff to carry out their operational goals and the core business of providing care to the seriously ill and injured. The ED is also recognised as a high-risk area concerning service demand and the potential for human error. Few studies have considered the perceptions and experiences of ED staff (physicians, nurses, allied health professionals, and administration staff) regarding the practice of teamwork, especially in the Kingdom of Saudi Arabia (KSA), and few studies have been conducted in KSA to explore the teamwork practices in EDs. Aim: To explore teamwork practices from the perspectives and experiences of staff (physicians, nurses, allied health professionals, and administration staff) when interacting with each other in the admission areas of the ED in a public hospital in the Northern Borders region of the KSA. Method: This research used a qualitative case study design, drawing on three data collection methods: documentary review, semi-structured interviews (n=22) with physicians (n=6), nurses (n=10), allied health professionals (n=3), and administrative staff (n=3) and six non- participant direct observations. All data were analysed using Braun and Clarke's thematic analysis. Findings: The thematic analysis of the data yielded ten prominent themes, which were integral to understanding the staff's views and experiences with teamwork in the ED's admission areas. These themes revealed the barriers and the enablers experienced by the ED teams. The findings demonstrated that interdisciplinary teamwork is encouraged by a shared goal of patient care, reduced employee stress, and improved job satisfaction. In the ED, interdisciplinary collaboration was critical and functioned under a hierarchical structure, with a clear leader guiding decisions to achieve the best outcomes. However, barriers such as negative staff behaviours, staff shortages, and inadequate management support often hindered teamwork. Moreover, the study found that gender interactions and the high number of expatriates in the ED posed challenges such as discrimination and language barriers. In addition, the triage process, such as admitting non-urgent patients, contributed to overcrowding and overwhelmed the teams. The findings revealed that effective teamwork depends on effective communication, multitasking, stress management, and leadership skills. Finally, these findings were examined and compared with the four domains (relational, processual, organisational, and contextual) of Reeves et al.'s (2010) conceptual framework for understanding interprofessional teamwork. However, individual factors emerged as a new fifth domain that is not a part of the framework also played a critical role in interdisciplinary teamwork in the ED. Conclusion: Effective interdisciplinary teamwork is crucial in the ED in KSA due to the diverse cultural issues staff experience. Gender, language, social, and organisational issues can sometimes impose barriers to collaboration. Consequently, continuous teamwork training and support are necessary to improve the teamwork practices of the ED staff and ensure the provision of high-quality care to patients. The study’s findings provide practical insights for healthcare policymakers, hospital administrators, the KSA Vision 2030, and other countries seeking to optimise patient safety and quality of care by implementing effective teamwork practices in an ED setting.48 0Item Restricted IMPROVING ASPECT-BASED SENTIMENT ANALYSIS THROUGH LARGE LANGUAGE MODELS(Florida state university, 2024) Alanazi, Sami; Liu, XiuwenAspect-Based Sentiment Analysis (ABSA) is a crucial task in Natural Language Processing (NLP) that seeks to extract sentiments associated with specific aspects within text data. While traditional sentiment analysis offers a broad view, ABSA provides a fine-grained approach by identifying sentiments tied to particular aspects, enabling deeper insights into user opinions across diverse domains. Despite improvements in NLP, accurately capturing aspect-specific sentiments, especially in complex and multi-aspect sentences, remains challenging due to the nuanced dependencies and variations in sentiment expression. Additionally, languages with limited annotated datasets, such as Arabic, present further obstacles in ABSA. This dissertation addresses these challenges by proposing methodologies that enhance ABSA capabilities through large language models and transformer architectures. Three primary approaches are developed and evaluated: First, aspect-specific sentiment classification using GPT-4 with prompt engineering to improve few-shot learning and in-context classification; second, triplet extraction utilizing an encoder-decoder framework based on the T5 model, designed to capture aspect-opinion-sentiment associations effectively; and lastly, Aspect-Aware Conditional BERT, an extension of AraBERT, incorporating a customized attention mechanism to dynamically adjust focus based on target aspects, particularly improving ABSA in multi-aspect Arabic text. Our experimental results demonstrate that these proposed methods outperform current baselines across multiple datasets, particularly in improving sentiment accuracy and aspect relevance. This research contributes new model architectures and techniques that enhance ABSA for high-resource and low-resource languages, offering a scalable solution adaptable to various domains.47 0