Theses for Computer Science

Permanent URI for this collection

https://bspace.buid.ac.ae/handle/1234/1099

Browse

Now showing 1 - 20 of 41

Promoting Social Sustainability through Metaverse Usage in Higher Education: An Examination of Knowledge Management Factors and Students' Perceptions
(The British University in Dubai (BUiD), 2024-04) AL MESMARI, SALEIMAH MOBARAK; Dr Mostafa Al-Emran
The study mainly focuses on evaluating the current status of the of the metaverse technology application in the education in general and within the UAE Higher education will be presented. The main aim of this research is to provide a comprehensive model for understanding the actual use of metaverse technology and to explore the subsequent impact on social sustainability within UAE HEIs.The objectives of this research are to review the existing literature on metaverse technology in higher education, identifying key factors that have been utilized to understand its adoption and usage. To develop an integrated research model that combines the TPB and T-EESST theories with knowledge management factors, privacy and security concerns, and metaverse-related factors. To empirically validate the developed integrated model using a hybrid Structural Equation Modelling and Artificial Neural Network (SEM-ANN) approach, ensuring a robust analysis of the relationships between the identified factors and students' actual usage behaviour of metaverse platforms in educational settings. This study has selected the quantitative technique over the other research approaches for the two main reasons. The first reason is because of the nature of the study. In this study, the theories have been tested by formulating precise hypotheses, and then gathered numerical data to prove or disprove those hypotheses. Statistical methods are utilised in order to do the analysis on the collected data. The findings of PLS-SEM show that factors such as attitude, knowledge acquisition, sharing, application, protection, immersion, presence, perceived security, and privacy play crucial roles in the usage of a system or technology. However, subjective norm, perceived behavioural control, and knowledge storage capabilities are not significant determinants of use. Importantly, the use of Metaverse is found to have a substantial positive impact on social sustainability, highlighting the importance of these factors in promoting sustainable social outcomes through Metaverse usage. The ANN analysis results align with the PLS-SEM findings, confirming the significant impact of factors like Knowledge Sharing, Knowledge Application, Presence, and Perceived Security on Metaverse use and its contribution to social sustainability. In this study, a number of practical implications offered that should be taken into account by metaverse platform decision-makers, administrators, developers, and educators in the field of higher education. The findings of this research provide valuable insights for developers and practitioners in the development and design of metaverse platforms for higher educational institutions, as they could examine the possible knowledge management aspects identified in this research. hese studies uncover distinct challenges and opportunities, providing a deeper understanding of the technology's applications and implications in diverse contexts.
A Decision Modelling Approach for Perceived and Engaged Immersive Visual Interactive Applications on the Metaverse
(The British University in Dubai (BUiD), 2024-09) IBRAHIM, SAMAR; Dr Mostafa Al Emran
Many researchers have developed and implemented various immersive visual interactive applications that incorporate visualization and interaction techniques, support the visual construct and enhance exploration in the Metaverse. Benchmarking these applications is critical due to the absence of an ideal model, particularly given their varied development criteria. The multiple evaluations, significance, and the data variability of these criteria among various applications present challenges that necessitate using specific multicriteria decision-making (MCDM) methods. This study introduces a new decision modelling approach that extends the fuzzy-weighted zero inconsistency (FWZIC) method with the new Interval-Valued Fermatean Neutrosophic fuzzy sets (IVFN) and combines it with the extended ranking method, Više kriterijumska Optimization I compromise Rešenje (VIKOR) and Gray relational analysis (GRA). The modelling approach benchmarks the most perceived immersive visual interactive applications using the extended VIKOR-GRA method based on a decision matrix of 18 criteria and 29 alternatives for the four visualization scenarios. The new extended IVFN-FWZIC method weighs the criteria for benchmarking the optimal applications based on experts’ judgments. The results illustrate the effectiveness of the modelling approach and indicate that "spatial" and “haptics” are the most significant criteria in evaluating immersive visual interactive applications. The robustness and reliability of the results were validated and evaluated using sensitivity analysis, comparative analysis, and a systematic ranking process. This study offers a vision for decision-makers and stakeholders, contributing to improving visualization and exploration of the Metaverse.
Developing a Digital Trust Framework between Government and Private Sector in the Financial Sector
(The British University in Dubai (BUiD), 2023-07) AL SHAMSI, SUAAD SALEM YOUSEF; Professor Sherief Abdallah
COVID-19 accelerated the transformation to a digital society. It created a necessity for an ecosystem trusted by all parties to avail services, which requires identity to be verified and information to be exchanged. The trusted digital identity of individuals and organizations is a fundamental block in such an ecosystem. Services in critical domains such as finance, health and travel heavily rely on identity verification. This process is known as “Know Your Customer” (KYC). Despite the need for a digital service offering, regulators mandate the physical verification of individuals and their documents. This process is usually lengthy, inefficient, costly and inconvenient to the customer. It mandates the customer to visit each service provider at least once to conduct customer identification and verification prior to availing the service or obtaining digital credentials. Regulators requirements need to be identified and addressed in a digital mean which is accepted and trusted by all involved parties, service providers, customers, and document issuers, in order to transform KYC process to a digital process which is trusted and efficient. The physical identity of individuals must be mapped to a digital identity that can be trusted and verified by all service providers. The physical documents of individuals representing their identity or any claim they make about themselves or their ownership must be transformed into trusted and verifiable digital credentials. This would enable an end to end digital service offering without the need for customer to physically visit service providers for customer identification and verification. This paper develop a Digital Trust Framework between the government and private sector in the United Arab Emirates (UAE) that aims to enable service providers to offer an end to end digital service through the development of a unified trusted digital identity and credentials that eliminate the need for the initial physical identification and verification, and imporve the continuous KYC monitoring of the customer therefore improving the efficiency and the user experience, reducing the cost, and increasing the transparency of the KYC processes. The framework uses decentralized identity and verifiable credentials (VCs) as a layer on top of blockchain technology to address the challenges faced in KYC processes while adopting Privacy-by-Design principles. Hence, this research paper follows a design science research methodology (DSRM) as research method to develop the framework. The focus is on the financial sector in the UAE. The DSRM research method integrates existing theoretical knowledge and insights from industry practitioners obtained through semi-structured expert interviews. The proposed framework to imporve the KYC process was successfully implemented as prototype system and it was evaluated through a case study evaluation with real users to verify its feasibility. The results of the demonstration were used to evaluate the prototype system through different performance metrics. The results of the evaluation showed satisfactory performance across various metrics indicating the system stability and reliability.
Leveraging Deep Learning and Word Embeddings to Detect Dish Names in Consumer Reviews
(The British University in Dubai (BUiD), 2024-02) ABOKHASHAN, DEENA YOUNIS; Professor Sherief Abdallah
Named Entity Recognition (NER) is crucial for extracting entities from unstructured text, offering significant insights for businesses through customer review analysis. This study fills a gap in recognizing dish names from customer reviews, as existing literature mainly addresses food entity recognition in recipe datasets and lacks annotated datasets for this specific NER task. Domain adaptation and deep learning approaches like BiGRUs and CNNs remain underexplored. The research proposes a deep learning NER framework to accurately identify dish names in customer reviews with efficient computational resource use. In addition to the existing dataset, MenuNER dataset, an annotated dataset, ReviewsDB, was created from Yelp reviews for evaluation. Initial experiments revealed a notable performance drop in domain adaptation from food names in recipe datasets to dish names in reviews, with the F1-score nearly 50% lower. A comparative analysis of 53 deep learning models using various word embeddings, including Glove, Word2Vec, and Bert variants, showed that a simple architecture with a single-layer Bidirectional Gated Recurrent Unit (BiGRU) and Conditional Random Field (CRF) layer achieved the best performance, with an F1-score of 93.07% using glove-twitter-100 embeddings in the MenuNER dataset. Additionally, a two-layer BiGRU with a CNN and CRF achieved an F1-score of 82.40% on the ReviewsDB dataset. The study attributes performance differences to variability in annotation lengths and the broader range of terms in ReviewsDB. In conclusion, the proposed NER framework, leveraging pre-trained embeddings, provides a valuable tool for the food industry to analyze customer feedback and enhance customer satisfaction.
Understanding the factors affecting users’ acceptance of smart government mobile applications using a hybrid SEM-ANN approach
(The British University in Dubai (BUiD), 2023-02) ZARE, AISHA ALI; Professor Sherief Abdullah
Abstract In recent years, smart cities have received significant attention as solutions to the complex problems modern communities face. The popularity and demand for smartphones and applications continue to increase, and smart governments are using innovative applications as essential tools to boost the effectiveness and efficiency of their services and to achieve transformation from an electronic government (e-government) to a mobile government (m-government). However, only a few scholars have explored those factors affecting the use of m-government innovative applications in developing countries. Therefore, the primary aim of this research is to determine those variables that influence the users’ actual use of m-government innovative applications and to examine their relationships. To do so, this study extends the Technology Acceptance Model (TAM) with trust in the internet, trust in government, cultural influence, service quality, and public awareness. The proposed theoretical model is evaluated based on data collected from 146 m-government service users using questionnaire surveys and is analyzed through a two-stage analytical technique that includes PLS-SEM and Artificial Neural Network (ANN). Among the 15 hypothesized associations in the research model, the correlations between variables support nine hypotheses, suggesting that these variables influence users’ adoption of smart government mobile applications. The results of this study reveal the positive impact of service quality, cultural influence, and public awareness on users’ perceived ease of use of smart government applications. The users’ perceived usefulness of smart government applications is also positively influenced by service quality and cultural influence. In contrast, their perceived ease of use and usefulness improves their attitude toward using these applications. The most important factor is perceived ease of use, affecting users’ intention and use of government smart applications. The most potent predictor of usage intention is attitude toward using smart applications. This study's key findings fill the literature gap and determine the most influential factors that impact end users, the state-of-the-art TAM model, and the effectiveness of the research model. These findings are expected to improve the current theorization of mobile technology elements that influence user acceptability and utilization of competent government mobile applications. This research also offers practical suggestions to m-government decision-makers and developers for improving the degree of actual use of their apps and services. Policymakers and practitioners in the smart government of Dubai can emphasize those factors that enhance the adoption of smart government mobile applications. Keywords: Smart cities, Mobile government (m-government), E-government, Technology Acceptance Model (TAM), Trust in the internet, Trust in government, Cultural influence, Service quality, Public awareness, User adoption, Smart government applications, Perceived ease of use, Perceived usefulness
Examining the Factors Influencing Autonomous Vehicle Use and its Impact on Environmental Sustainability in UAE using a Hybrid SEM-ANN Approach
(The British University in Dubai (BUiD), 2024-05) AL MANSOORI, SAEED HUSSAIN ALI; Dr Mostafa Al-Emran,
Over the past few years, the evolution of transportation towards autonomous vehicles has attracted substantial interest, signifying a transformative shift in the realm of mobility. Nevertheless, AVs adoption is a complex process, influenced by cutting-edge technological advancements, societal acceptance, regulatory frameworks, and potential environmental impacts, all of which are critical in shaping their integration into modern transportation systems. To address the gap in the existing research of this domain, a comprehensive systematic review was conducted, focusing on AVs adoption through the lens of various information system (IS) models and theoretical frameworks. The systematic review analysed empirical studies published in the timeframe from January 2013 to January 2023, focusing on AVs adoption. Out of 3,532 articles, 71 were shortlisted for in-depth analysis according to specific inclusion criteria. In addition, this study constructs a novel theoretical model based on the integration of the Protection Motivation Theory (PMT), the Behavioural Reasoning Theory (BRT), and the updated IS success model variables to investigate the factors influencing AVs adoption and its impact on environmental sustainability. Subsequently, the proposed model was validated using data obtained from an online survey involving 495 individuals in the UAE, who either own or have had experience using AVs. The proposed model underwent empirical validation by applying a hybrid Structural Equation Modelling-Artificial Neural Network (SEM-ANN). The results of hypothesis testing provided strong support for the majority of the hypotheses derived from the proposed model (i.e., out of 12 hypotheses, 9 were supported).
Enhancing Arabic Offensive Tweet Classification: An Ensemble Approach Integrating AraBERT, Neural Networks, and LSTM Models
(The British University in Dubai (BUiD), 2023-10) WAHDAN, AHLAM MOHAMMAD; Professor Khaled Shaalan; Dr Mostafa AL-Emran
This thesis addresses the crucial research problem of accurate detection and moderation of offensive language in Arabic text, considering the intricacies posed by the language's complex morphology, dialectal variations, orthographic ambiguity, orthographic noise, limited linguistic resources, and the necessity for comprehensive coverage of offensive language expressions. The research objectives are delineated through four key research questions. Firstly, the study aims to identify the existing research gaps in Arabic Text Classification (ATC) through an extensive and rigorous systematic literature review. The study adopts a scholarly and formal approach, aiming to identify the specific areas within ATC research that lack comprehensive exploration or exhibit inadequacies in existing knowledge. This endeavor is grounded in the rigorous analysis and synthesis of relevant academic literature, ensuring a meticulous examination of the current state of research in ATC. Secondly, it investigates the effects of employing novel pre-processing methods on the performance of Arabic Text Classification. Thirdly, the research endeavors to determine the most effective model for enhancing the accuracy of Arabic offensive text classification by introducing a novel approach using pre-trained models; AraBERT model in conjunction with fully connected neural networks (NN) and long short-term memory (LSTM) networks. Finally, the study evaluates the proposed model's ability to classify Arabic offensive text effectively. The research methodology consists of two integral parts, comprising dataset description, the proposed framework. The dataset description provides insights into the two datasets utilized, namely OSACT and SEMEval. The framework elucidates the proposed model, which leverages a combination of pretrained models and neural networks, thereby achieving a high level of effectiveness in classifying Arabic offensive text. The model's performance is meticulously assessed using various evaluation metrics, including accuracy and F1-macro score, and is compared against other classifier models. The research findings demonstrate the superiority of the proposed model over the baseline AraBERT model, with the proposed model achieving an accuracy of 0.870 compared to the baseline accuracy of 0.820, along with an F1-score of 0.853 compared to the baseline's 0.800. This emphasizes the model's exceptional capacity to accurately identify offensive content in Arabic text. The implications of this research extend to diverse domains and stakeholders, encompassing decision makers, developers, and policy makers. The insights garnered from the study can be instrumental in making informed decisions pertaining to the integration of Arabic text classification systems in various operational settings. By comprehending the proposed model's performance and efficacy, decision makers can assess its potential impact on optimizing processes such as information retrieval, content filtering, and sentiment analysis in Arabic text. In conclusion, this thesis contributes significantly to the existing literature by addressing the complexities associated with offensive language identification in Arabic text and introducing an innovative approach that integrates pretrained models with deep learning techniques and neural networks. The demonstrated effectiveness and superior performance of the proposed model underscore its potential for practical implementation in real-world scenarios, thereby bolstering the field of Arabic offensive text classification.
Sentiment Analysis in Social Media: A Deep Learning Framework for Analyzing Public Sentiment toward Policies during Natural Disasters
(The British University in Dubai (BUiD), 2023-07) FOOLAD, MOHAMED ABDULLA; Professor Sherief Abdallah
The primary aim of this study is to produce and present a robust, reliable, and accurate framework to measure and analyze public sentiments for decision-makers and policies legislators while managing and administrating the spread of a pandemic or a natural disaster in general. The study utilizes COVID-19 as an actual case study to evaluate the proposed framework, and tests relevant hypotheses. The study used word embedding as a feature vector and then tested the accuracy of different models such as LSTM, Transformers, Logistic regression, Random Forest Classifier, KNN, and Multinomial Naïve Bayes. This work used a deep learning model –LSTM model on the collected tweets to classify the emotions and reactions. The models applied in the classification of COVID-19 global data performed differently with regard to accuracy and precision. Transformers emerged as the most accurate model, with a precision score of (90.18%) and an accuracy of (90.17%). LSTM was also superior with regard to the accuracy, with an (88.45%) precision score and an accuracy of (88.44%). The Random Forest classifier performed fairly, with a (52.43%) precision score and (51.26%). The other models performed poorly in predicting sentiments; for instance, Multinomial Naïve Bayes recorded (43.54%) precision and (41.23%) prediction accuracy, the logistic regression model recorded a (43.54%) precision and (41.23%) prediction accuracy, and the KNN model had (41.23%) precision and (41.11%) prediction accuracy. Overall, LSTM was established as the most suitable model for the selected dataset since it was able to fit the dataset and can be generalized to the current and new datasets, unlike transformers.
Exploring the Factors Affecting Chatbot Use in Higher Education and Its Impact on Social Sustainability Using a Hybrid SEM-ANN Approach
(The British University in Dubai (BUiD), 2023-06) ALNUAIMI, ABDULLA ALSHARHAN; Professor Khaled Shaalan
Previous studies have identified various drivers of chatbot adoption through different technology adoption theories. However, these studies have not been thoroughly reviewed and synthesized. Most studies have focused on examining the intention of using chatbots, with limited investigations on actual use and sustainable intention. This thesis explores the factors influencing the sustainable adoption of chatbots in higher education, specifically focusing on social sustainability. It aims to develop an integrated model that comprehensively examines the influencing factors, moderating effects, and organizational dynamics that shape chatbot adoption. The research begins with a comprehensive literature review to identify factors that generally influence chatbot adoption. These factors encompass technological, individual, organizational, and contextual dimensions. Building upon this literature review, the conceptual model is developed, addressing the research gaps in the existing literature and focusing on chatbot acceptance in the higher education sector. The systematic review includes an analysis of empirical studies published between 2016 and September 2022, resulting in 219 eligible studies out of 3,942 reviewed. The main findings reveal that the Technology Acceptance Model (TAM), Social Presence Theory (SPT), and Computers as Social Actors (CASA) are the prevailing theories explaining chatbot adoption. Anthropomorphism is the most examined external factor, followed by trust, enjoyment, and interactivity. The conceptual framework integrates established theories such as Task-Technology Fit (TTF), Source Credibility Theory, Social Presence Theory (SPT), and additional factors specific to chatbot adoption in higher education. A quantitative survey is administered to a sample of 341 individuals and students from the higher education sector in the UAE, capturing diverse perspectives on chatbot adoption. The model is then validated using advanced analytical techniques, including Partial Least Squares Structural Equation Modelling (PLS-SEM) and Artificial Neural Networks (ANN). The findings contribute to understanding chatbot adoption in higher education, providing insights into key drivers, challenges, and implications for social sustainability. The study's findings highlight factors such as task-technology fit, credibility, and social presence that significantly influence the intention to sustainably use chatbots in higher education. The sensitivity analysis reveals the importance of social presence, followed by credibility and task-technology fit, in influencing chatbot use. The chatbots’ technological characteristics have a greater impact than the task characteristics, and visual cues are perceived as more important than invisible and verbal cues for chatbot social presence. Trustworthiness is the most significant factor impacting credibility, followed by ease of use, tailoring, and commercial implications. However, Expertise, Real-World Feel, and Amateurism do not significantly impact credibility. These results contribute to developing acceptance models that can guide the design, implementation, and evaluation of chatbot initiatives in higher education institutions, fostering social sustainability. The theoretical contributions lie in developing an integrated model that extends existing theories to the context of chatbot adoption in higher education. The model provides a comprehensive understanding of the influencing factors and their interrelationships, offering a valuable framework for future research. From a practical perspective, the findings assist higher education institutions in strategically implementing and managing chatbots. The insights gained from this study can guide the development of effective strategies to promote chatbot acceptance, address privacy concerns, and leverage chatbots' potential for enhancing social sustainability in higher education settings.
Detection of Depression in Arabic Social Media: A Comparison of Traditional and Modern Machine Learning Algorithms
(The British University in Dubai (BUiD), 2023-12) ALSHEHHI, OMAR KHALID HAMAD; Professor Sherief Abdallah
This study aims to address the research gap in detecting depression from Arabic tweets using the PHQ-9 scale as a framework. The dataset collected was a set of 200,000 tweets from around 20,000 users. A team of psychologists and assistants used a user-based approach to label users as either depressed or not. The data labelling and annotation process involved a user-based evaluation of the tweets to label users as either depressed or not, based on the two target variables of depressed_binary and depressed_multi. Users with scores between 0 and 6 were categorized as not depressed in the depressed_binary variable, while those with scores above six were classified as depressed. For the depressed_multi variable, users with scores ranging from 0 to 2 were labelled as not depressed, scores from 3 to 6 indicated mild depression, scores from 7 to 9 indicated moderate depression and scores of 10 or above represented high depression. Four machine learning models were employed in this study: HGB (Histogram Gradient Boost), GRU (Gated Recurrent Units), LSTM (Long Short-Term Memory), and SVM (Support Vector Machines). The findings revealed that the older models exhibited strong performance in binary classification, while the new models demonstrated competitive results. Future research should focus on exploring and developing newer deep learning models, such as HGB and GRU models, to enhance the accuracy and performance of depression detection in Arabic tweets. Future studies should also investigate strategies to account for the influence of different Arabic dialects and incorporate Arabic colloquialisms in depression detection models.
Sentiment Analysis of the Emirati Dialect text using Ensemble Stacking Deep Learning Models
(The British University in Dubai (BUiD), 2023-03) AL SHAMSI, ARWA AHMED; Professor Sherief Abdallah
The study of thoughts, feelings, judgments, values, attitudes, and emotions regarding goods, services, organizations, persons, tasks, occasions, titles, and their attributes is known as sentiment analysis and it involves a polarity classification task for recognizing positive, negative, or neutral text to quantify what individuals believe using textual qualitative data. The rise of social media platforms provided an excellent source for sentiment analysis data. People use these platforms for various reasons, ranging from sharing their opinions and thoughts to gaining knowledge. Twitter, Instagram, and Facebook are examples of social media platforms. As more users join social media platforms, the amount of data that is generated online continues to grow at an accelerating pace. Most of the previous research that studied sentiment analysis for the Arabic language focused on Modern Standard Arabic and Egyptian, Saudi, Algerian, Jordanian, Tunisian, and Levantine Dialects. However, to our knowledge, no study involved employing deep learning models to conduct sentiment analysis on the Emirati dialect texts. Dialects are the informal form of a language. Each country of the Arab world has its own Dialect, and each dialect may have several sub-dialects. The main objective of this study is to develop a deep learning model that outperforms the state-of-the-art for Sentiment Analysis of the Emirati Dialect. Toward this objective, I first conducted a systematic review to identify the research gaps in the existing literature and investigate the available constructed resources for Arabic dialects and the used approaches for sentiment analysis of Arabic dialects. The systematic review focused on empirical research on the subject of Sentiment Analysis of the Arabic Dialect that was released between January 2015 and January 2021. Through the analysis, I found, with the exception of a few articles that investigated Saudi, Levantine, Jordanian, Algerian, Tunisian, and Egyptian dialects, researchers rarely specified the dialect type in their papers; instead, it was mostly mentioned (MSA and Arabic Dialects). Emirati Dialect has not been explored for Sentiment Analysis purposes. The sizes of most datasets of previous research were between 10,000 and 50,000. Moreover, the Twitter platform was the most popular online platform for constructing Arabic datasets. Most of the studies evaluated basic Machine Learning approaches for Sentiment Analysis of Arabic Dialects. My research aims to fill these identified gaps as I detail below. Since Instagram is one of the most popular social media platforms in UAE, I constructed a dataset of the Emirati dialect from the Instagram platform. My dataset consists of 216,000 posts, of which 70,000 posts were manually annotated by three human annotators. Each post is annotated into (Positive/ Negative/Neutral), and it is further annotated into (Emirati Dialect/ Arabic Dialect/ MSA). In order for the dataset to be used as a benchmark, the inter-annotator agreement (IAA) was measured using Fleiss's Kappa coefficient. The findings reveal that the overall Fleiss Kappa coefficient is = 0.93, indicating an almost-perfect agreement amongst the three annotators. Once the dataset was constructed and validated, I then conducted a performance evaluation and comparison of various basic Machine Learning algorithms, Deep Learning models, and stacking deep learning models on different datasets of Sentiment Analysis of Arabic Dialects. For the basic machine learning algorithms, LR, NB, SVM, RF, DT, MLP, AdaBoost, GBoost, and an ensemble model of machine learning classifiers were used. For deep-learning models, CNN, Bi-LSTM, Bi-GRU, as well as Hybrid deep-learning models were used for Sentiment Analysis. In order to improve performance further, I have proposed three ensemble-stacking deep-learning models with meta-learner layers of classifiers. The first stacking deep learning model combined 2 of the used deep learning models that produced the best results in terms of accuracy, the second stacking deep learning model combined 4 of the used deep learning models that produced the best results in terms of accuracy, and the final stacking deep-learning model combined all the trained deep learning models in this research. The proposed ensemble stacking model was evaluated using three datasets: the ESAAD Emirati Sentiment Analysis Annotated Dataset (which is one of this thesis contributions), and two other benchmark datasets (A Twitter-based Benchmark Arabic Sentiment Analysis Dataset ASAD and Arabic Company Reviews dataset). Experimental results show that my proposed ensemble stacking model outperformed existing deep learning models and achieved an accuracy of 95.54% for the ESAAD dataset, 96.71% for the benchmark ASAD, and 96.65% for the Arabic Company Reviews Dataset.
Cyberbullying Detection in Arabic Text using Deep Learning
(The British University in Dubai (BUiD), 2023-03) ALBAYARI, REEM RAMADAN SA’ID; Professor Sherief Abdallah
In the new era of digital communications, cyberbullying is a significant concern for society. Cyberbullying involves the use of communication technology and data, including messages, photographs, and videos, to undertake aggressive negative actions to harm others. This practice has spread substantially due to rapid technological development and has gained significant attention in several domains involving data exchange, such as e-commerce, digital marketing, social media platforms, and others. Cyberbullying can negatively impact stakeholders, and can vary from psychological to pathological, such as self-isolation, depression, and anxiety potentially leading to suicide. Hence, detecting any act of cyberbullying in an automated manner will be helpful for stakeholders to prevent any unfortunate results from the victim’s perspective. If conducted automatically, rather than relying on human moderators, the process will be faster, enabling the early detection of cyberbullying before severe harm is caused. Data-driven approaches, such as machine learning (ML), particularly deep learning (DL), have shown promising results. DL approaches provide highly accurate predictive models for detecting cyberbullying. The first contribution of this thesis is conducting an in-depth meta-analysis of existing evaluation methods, classification techniques, and datasets related to ML for cyberbullying problems. The meta-analysis shows that ML approaches, particularly DL, have not been extensively studied for the Arabic text classification of cyberbullying. A potential reason for this research gap is the lack of Arabic-language repositories focusing on cyberbullying despite the large amount of Arabic text that can be extracted from Arabic social media platforms besides e-commerce and mobile applications. Consequently, I have designed and built a new Arabic text repository, the largest available, that can serve me and others in investigating various classifiers to deal with the issue of detecting cyberbullying. This repository contains 200,000 comments, 46,898 of which were annotated by three human annotators. First, the comments were classified as (positive/negative/neutral), and then the negative comments were further classified into two categories based on their level of negativity (toxic, bullying). The dialect for each comment was also added. This gives the dataset an advantage since it can be used for other purposes such as sentiment analysis and dialect identification, not just for cyberbullying detection. For the dataset to be regarded as a benchmark, Fleiss’s Kappa metric was adopted to measure the inter-annotator agreement (IAA), and the results show that the total Fleiss Kappa coefficient is = 0.869 with a p-value of 10-3, indicating near-perfect agreement among the three annotators. The application of DL to cyberbullying detection problems within Arabic text classification can be considered a novel approach due to the complexity of the problem and the tedious process involved, besides the scarcity of relevant research studies. Therefore, this study aims to evaluate several versions of Recurrent Neural Networks (RNNs) and Feedforward Neural Networks (FNNs) for detecting cyberbullying in the Arabic language. Although these algorithms are widely used in text classification and outperform the performance of classical classifiers, many have been extensively investigated in other domains such as sentiment analysis and dialect identification, as well as cyberbullying detection in English text. Hence, a comprehensive study focusing on Arabic cyberbullying can fill this gap in research. In this study, I conduct a performance evaluation and comparison for various DL algorithms (LSTM, GRU, LSTM-ATT, CNN-BLSTM, CNN-LSTM, CNN-BILSTM-LSTM, and LSTM-TCN) on different datasets of Arabic cyberbullying to obtain more precise and dependable findings. As a result of the models’ evaluation, a hybrid DL model is proposed that combines the best characteristics of the baseline models CNN, BLSTM and GRU for identifying cyberbullying. The proposed hybrid model improves the accuracy of all the studied datasets and can be integrated into different social media sites to automatically detect cyberbullying from Arabic social media posts. It has the potential to significantly reduce cyberbullying. Other results, related implications, and limitations, along with future research are also clarified and discussed.
Governmental Data Analytics: An Agile Framework Development and Real-World Data Analytics Case Studies
(The British University in Dubai (BUiD), 2023-03) QADADEH, WAFA; Professor Sherief Abdallah
A Decision Modelling Approach for Security Modules of Delegation Methods in Mobile Cloud Computing using Probabilistic Interval Neutrosophic Hesitant Fuzzy Set
(The British University in Dubai (BUiD), 2023-03) AL HANTOOBI, SENDEYAH
Mobile Cloud Computing (MCC) has become a pervasive technology that offers on-demand, flexible, and scalable computing resources to mobile devices. However, the security issues associated with MCC have become a major concern for users and organizations, leading to the development of various Security Modules. These modules typically use delegation methods that involve the transfer of data or operations from mobile devices to the cloud to perform the task at the best performance and security levels. Despite extensive attempts to design secure security modules of delegation in Mobile Cloud Computing (MCC), none of the existing modules possess all the necessary development attributes. Our analysis indicates that previous studies have not used security development attributes as evaluation criteria to compare and assess the available Security modules of delegation methods in MCC. However, Modeling these modules is critical and poses significant challenges in selecting the most secure security module. du to multicriteria, importance of data and data variation. To address this issue, this study proposes a Decision Modelling Approach for Security Modules of Delegation Methods in Mobile Cloud Computing using multi-criteria decision-making (MCDM) methods. The proposed approach involves the integration of Evaluation based on Distance from Average Solution (EDAS) method with fuzzy weighted with zero inconsistency (FWZIC) under Probabilistic Interval Neutrosophic Hesitant Fuzzy Set (PINHFS) environment. . The framework presented in this study involves two primary stages : the construction of decision matrices for Security Modules of delegation methods in MCC and the application of the PINHFS-FWZIC method to determine the weight of the security evaluation criteria. The EDAS method is then employed to modeling the Security Modules of delegation methods in MCC based on the formulated decision matrices and criteria weight. The validation and evaluation of the proposed framework were conducted through model validation and decision evaluation procedures. Model validation involved sensitivity analysis and systematic ranking procedures . The benchmarking checklist was used to compare the results of the proposed framework with the existing approaches. Based on the findings, it can be concluded that the proposed framework can efficiently weight the security criteria and successfully rank Security Modules of delegation methods in MCC. The PINHFS-FWZIC method effectively handled the uncertainty and hesitancy of the decision-makers in assigning weights to the evaluation criteria. Overall, the proposed framework provides a useful benchmark for evaluating other Security Modules of delegation methods in MCC. It can aid decision-makers in selecting the most secure MCC delegation method system by providing a comprehensive evaluation and Modeling of the available Security Modules.
Factors Affecting Cybersecurity Behaviour in the Metaverse: A Hybrid SEM-ANN Approach Based on Deep Learning
(The British University in Dubai (BUiD), 2023-01) ALMANSOORI, AFRAH
Cybersecurity procedures and policies are prevalent countermeasures for protecting organizations from cybercrimes and security incidents. However, without considering human behaviours, implementing these countermeasures will remain useless. Cybersecurity behaviour has gained much attention in recent years. However, little is known concerning the factors that influence the cybersecurity behaviour of Metaverse users. Consequently, this research has three key objectives. A comprehensive systematic review is steered to address research gaps in the current literature on cybersecurity behaviour via the lens of information system models and theories to identify the most prevalent factors, theoretical models, technologies and services, and participants. The systematic review identified 2,210 empirical studies published on cybersecurity behaviour between 2012 and 2021. In line with the existing gaps found in the literature, this research, therefore, develops an integrated model based on extracting constructs from the Protection Motivation Theory (PMT), Health Belief Model (HBM), and Theory of Interpersonal Behaviour (TIB). An external factor, “trust”, is also incorporated in the model to understand better the factors affecting the cybersecurity behaviour in the Metaverse. The developed model was then evaluated based on survey responses from 531 Metaverse users in the United Arab Emirates who used the Metaverse for personal or professional purposes. The empirical data were analysed using a deep learning-based hybrid Structural Equation Modeling (SEM) and Artificial Neural Network (ANN) Approach. The integrated model explained 66.1% of the total variance in cybersecurity behaviour. The hypotheses testing results reinforced most of the suggested hypotheses in the developed model. The sensitivity analysis results for the ANN model revealed that “cues to action” have the most considerable importance in understanding cybersecurity behaviour in the Metaverse, with 97.8% normalized importance, followed by habit (69.7%), perceived vulnerability (69.6%), self-efficacy (40.9%), and trust (27.2%). Theoretically, integrating the PMT, HBM, and TIB along with the external factor, “trust”, is believed to add a significant value to validating the three theories in general and the cybersecurity behaviour in specific. Practically, understanding the impact of security factors would assist in understanding the effect of security incidents on cybersecurity behaviour in the Metaverse. Furthermore, policymakers and regulators should pay attention to and analyse the present data privacy policies and legislation to create specific policies and regulations for using the Metaverse. Moreover, cybersecurity companies, system analysts, and developers can use the insights from the essential factors as a form of a lesson to the refinement of presently implemented solutions as well as the anticipation of new future technological advancements.
Examining the Factors affecting the sustainability of blockchain technology in higher education using a hybrid SEM-ANN approach
(The British University in Dubai (BUiD), 2022-10) AL SHAMSI, MOHAMMED SALEM MATAR
Blockchain technology has received considerable attention during the last few years. Blockchain technology permits the creation of a distributed record of a decentralized digital event in which data and related transactions are not controlled by a third party. However, little is known concerning what aﬀects their sustainable use for educational purposes. Consequently, there are three primary goals for this research. A systematic review is conducted to address the research gaps in the literature on Blockchain adoption from the lens of information system (IS) models and theories. The systematic review included empirical studies published between January 2010 and December 2021 on the topic of blockchain adoption. Among the 918 articles, 30 articles were critically analyzed based on the inclusion criteria. This research, therefore, develops a theoretical model based on extracting constructs from the protection motivation theory (PMT) and expectation conﬁrmation model (ECM) to understand the sustainable use of blockchain in higher education. The developed model was then tested based on data collected through an online survey from 374 university students in the UAE who used blockchain technology for educational purposes. The model was empirically validated using a hybrid structural equation modelling-artiﬁcial neural network (SEM-ANN) approach. The hypotheses testing results reinforced most of the suggested hypotheses in the developed model. The sensitivity analysis results for model 1 revealed that satisfaction has the most considerable eﬀect on the sustainable use of blockchain technology with 100% normalized importance, followed by perceived usefulness (58.8%), perceived severity (12.1%), and response cost (9.2%). Besides, the sensitivity analysis for model 2 showed that perceived usefulness has the most considerable effect on the sustainable use of blockchain technology with 100% normalized importance. However, expectation confirmation has 29.2%. Theoretically, integrating the PMT and ECM will add significant value to the validation of the two theories in general and the blockchain in specific. Practically, understanding the impact of security factors would assist in understanding the effect of security incidents on the sustainable use of technology in higher educational institutions. In addition, governments, academia, businesses, and individuals often tend to share resources over a distributed ledger secured by means of cryptography. Blockchain technology also helps in facilitating the traceable, secure, and verifiable exchange of educational data across institutions effectively. Methodologically, the use of SEM-ANN in validating such theoretical models is rarely used, and hence, it would add value to the existing literature by measuring the non-linear relationships among the factors.
Examining the factors affecting users’ cybersecurity behaviour in mobile payment contactless technologies: A hybrid SEM-ANN approach
(The British University in Dubai (BUiD), 2022-12) ZAINAL, HANA YOUSUF
Modern businesses have adopted a plethora of advancements due to the emergence of digital transformation where the recent lifestyle requires to have a robust platform to process all the communication in an effective and secure way. Advanced technologies adopted by the society purely concentrated on the software and cybersecurity enhancements from the software perspective ignoring the fact that a human behaviour can be the most major and severe threat for these advanced technologies. The adoption of any new technology in the society has been turned into a must rather than a complimentary due to the rapid growth of the technology advancements and people curiosity to investigate new trends. Technology adoption depends on the behaviour of the end users in dealing with the technology. Mobile payment contactless technologies are one of the fast-growing technologies specially after the appearance of the COVID-19 where all the payments and transactions are performed using RFID (Radio-Frequency Identification) or NFC (Near Field Communication) techniques. The individual’s cybersecurity behaviour is a major concern in these technologies. A quick review of literature showed that none of the studies discussed the individual’s cybersecurity behaviour toward this technology. Therefore, this research has three main objectives. First, conducting a systematic of systematic reviews on cybersecurity through a multidisciplinary perspective to address the research gaps in the existing literature of human cybersecurity behaviour. The systematic review of systematic review included empirical studies published between January 2015 and February 2022 on the topic of cybersecurity. A total of 794 studies have been reviewed, and 60 studies were found qualified and considered in the analysis. Through the analysis, little attention has been paid to cybersecurity behaviour in general and mobile payment contactless technologies in specific. Moreover, the existing literature discussed the utilization of mobile payment contactless technologies in terms of adoption and use in a certain criteria or environment and do not measure or explore the cybersecurity behaviour towards this technology. To bridge this literature gap, this research builds an integrated theoretical model by combining factors from the Protection Motivation Theory (PMT), Technology Threat Avoidance Theory (TTAT), Theory of Planned Behaviour (TPB), along with cybersecurity awareness to measure the cybersecurity behaviour in mobile payment contactless technologies. The model was then empirically validated using a hybrid Structural Equation Modelling-Artificial Neural Network (SEM-ANN) approach based on data collected through questionnaire surveys from mobile payment contactless technologies users in the UAE. The results indicated that the attitude, subjective norm, cybersecurity awareness, perceived behavioural control and response efficacy have a significant influence on cybersecurity behaviour. On the other side, the results indicated that perceived threat, response cost, and self-efficacy have insignificant influence on cybersecurity behaviour. The model explained 53% of the total variance in cybersecurity behaviour. Additionally, the ANN results showed that cybersecurity awareness is the most important factor affecting cybersecurity behaviour, with a normalized importance of 100%. Theoretical contributions, practical implications, and limitations and future research were also discussed.
A Framework to Implement Advanced Technologies in Higher Education Institutions: A Blockchain-Based Solution Case Study
(The British University in Dubai (BUiD), 2022-11) AL MANSOORI, SUAAD
The main drive behind this research is a statement made by Haugsbakken and Langseth (2019) that higher education institutions (HEIs) have a history of being slow and ineffective in adopting new technologies. The United Arab Emirates has encouraged HEIs to adopt new technologies to reinforce services. One of the main operations in HEIs is managing students’ credentials which requires HEIs to implement smart solutions rather than following traditional practices such as filling paper-based forms and getting approvals via emails. However, smart solutions, technologies, and advanced systems are associated with several challenges that might drive decision-makers away from replacing the traditional practices with new advances solutions in HEIs. This research proposes a general implementation framework that can guide decision-makers in HEI when implementing new solutions and technologies. The research also takes Blockchain Technology as a case study to customize the proposed framework and use it to implement Blockchain-based solutions. The researcher has conducted several qualitative and quantitative methods. First, highlighting challenges of implementing new solutions that were discussed in literature; with the focus on Blockchain as one of the most promising technologies. Second, building an online survey questionnaire in which HEIs stakeholders participated in to rank the challenges and provide further insights. Third, studying the case of a private university in The United Arab Emirates to highlight the issues of the current traditional credentials management practices and use the new proposed framework to implement a new Blockchain-based system in the same university. For the validation, the research was presented and published at two international conferences. It was also validated with the experts’ feedback and a live demo. This research does not focus on the technical aspects of Blockchain technology or any other technology but serves as a guideline to motivate decision-makers and stakeholders in HEIs to adopt new technologies.
Federal Multi Criteria Decision Making Framework in Distribution of Anti-SARS-CoV-2 Monoclonal Antibody to Eligible High-Risk Patients as Case Study
(The British University in Dubai (BUiD), 2022-06) ALSEREIDI, ABEER
No specific treatment was available for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) when the epidemic firstly broke out. The urgent need to end this unusual situation has resulted in many attempts to deal with SARS-CoV-2. In addition to several types of vaccinations that have been created, anti-SARS-CoV-2 monoclonal antibodies (mAbs) have added a new dimension to preventative and treatment efforts. This therapy also helps prevent severe symptoms for those at a high risk. Therefore, it is a promising treatment for mild-to-moderate SARS-CoV-2 cases. However, the availability of anti-SARS-CoV-2 mAb therapy is limited and leads to two main challenges. The first is the privacy challenge of selecting eligible patients from the distribution hospital networking, which requires data sharing; the second is the prioritisation of all eligible patients amongst the distribution hospitals according to dose availability. To our knowledge, no research has combined the federated fundamental approach with multicriteria decision-making methods for the treatment of SARS-COV-2, which indicates a research gap. This thesis presents a unique sequence processing methodology that distributes anti-SARS-CoV-2 mAbs to eligible high-risk patients with SARS-CoV-2 according to medical requirements by using a novel federated decision-making distributor (FDMD). A novel FDMD of anti-SARS-CoV-2 mAbs is proposed for eligible high-risk patients. FDMD is implemented on augmented data of 49,152 cases of patients with SARS-CoV-2 with mild and moderate symptoms. For proof of concept, three hospitals with 16 patients each are enrolled. The proposed FDMD is constructed from the two sides of claim sequencing: central federated server (CFS) and local machine (LM). The CFS includes five sequential phases synchronised with the LMs, namely, the preliminary criteria setting phase that determines the high-risk criteria, calculates their weights using the newly formulated interval-valued spherical fuzzy and hesitant 2-tuple fuzzy-weighted zero-inconsistency (IVSH2-FWZIC) and allocates their values. The subsequent phases are federation, dose availability confirmation, global prioritisation of eligible patients and alerting the hospitals with the patients most eligible for receiving the anti-SARS-CoV-2 mAbs according to dose availability. The LM independently performs all local prioritisation processes without sharing patients’ data using the provided criteria settings and federated parameters from the CFS via the proposed federated TOPSIS (F-TOPSIS). The sequential processing steps are coherently performed at both sides. The results are presented as follows: (1) The proposed FDMD efficiently and independently identifies the high-risk patients who are most eligible for receiving anti-SARS-CoV-2 mAbs at each local distribution hospital. The final decision at the CFS relies on the indexed patients’ score and dose availability without sharing the patients’ data. (2) The IVSH2-FWZIC effectively weights the high-risk criteria of patients with SARS-CoV-2. (3) The local and global prioritisation ranks of the F-TOPSIS for eligible patients are subjected to a systematic ranking validated by high correlation results across nine scenarios by altering the weights of the criteria. (4) A comparative analysis of the experimental results with a prior study confirms the effectiveness of the proposed FDMD. The study of the proposed FDMD implies that it has the benefits of centrally distributing anti-SARS-CoV-2 mAbs to high-risk patients prioritised according to their eligibility and dose availability. It also simultaneously protects their privacy and offers an effective cure to prevent progression to severe SARS-CoV-2, hospitalisation or death.
NOVEL STACKING CLASSIFICATION AND PREDICTION ALGORITHM BASED AMBIENT ASSISTED LIVING FOR ELDERLY
(The British University in Dubai (BUiD), 2022-09) JINESH, PADIKKAPPARAMBIL
The ageing of the population in developed nations necessitates the expansion of medical services, raising the cost of both economic and human resources. In this respect, Ambient Assisted Living (AAL) is a comparatively new information and communication technology (ICT) that delivers services. It acknowledges numerous products that help elderly and disabled people to live autonomously and enrich the quality of their lives. It also aids in the cost-cutting of hospital services. Various sensors and equipment are installed in the AAL context to collect a wide variety of data. Furthermore, AAL could be the motivating technique for the most recent care models by working as an adjunct. Research and development (R&D) projects and business activities in AAL and smart home contexts frequently emphasize the significance of technology. While ICTs promote health care, they have the potential to alleviate loneliness and social isolation among the elderly. They help to enhance, expand, and maintain the social interactions of the elderly while also increasing the individual's emotional well-being. The emergence of smart homes will help the elderly and the disabled live better lives. Over the past five years, the acceptance of wearable fall detection technologies has increased. These techniques involve calling for help in an emergency, such as falling or being immobile for long periods. Because falling is widespread among older adults, it can have serious health consequences. Falls can result in physical traumas such as fractures, head injuries, and severe decay. Falls will have a considerable impact on some populations, necessitating the development of better fall prevention and management solutions. Therefore, this thesis proposed a Novel Stacking Classification and Prediction (NSCP) algorithm based on AAL for the older people with Multi-strategy Combination based Feature Selection (MCFS) and Novel Clustering Aggregation (NCA) algorithms. The primary objective of this thesis is to identify the fall detection and prediction in older persons, such as 0 - no fall detected, 1 - person slipped/tripped / fall prediction, and 2 - definite fall. This study's dataset was sourced from the Kaggle machine learning repository, and it refers to data gathering from wearable IoT devices. The experimental outcomes demonstrate the proposed MCFS, NCA, and NSCP algorithms work more effectively than previous feature selection, clustering and classification algorithms, respectively, in terms of accuracy, sensitivity, specificity, precision, recall, f-measure and execution time. This thesis concludes with a discussion of future work to improve the proposed methodology and future research directions.

Browse

Recent Submissions