Journal of King Abdulaziz University: Computing and Information Technology Sciences

Task-Oriented Authoring Tool Using ChatGPT to Create Educational Textbooks

Miada Ahmeddeab Almasre, Alanoud Talal Subahi — Sun, 11 Aug 2024 00:00:00 +0000

The accelerated development of technology has led to the emergence of cutting-edge smart tools, such as artificial intelligence (AI) chatbots and machine learning algorithms, which possess substantial potential for improving learning and education. Conventional content creation tools frequently lack these sophisticated features, rendering the incorporation of AI, including OpenAI's ChatGPT, an appealing area to investigate. This study aims to assess the effectiveness, cognitive load, usability, and potential challenges of a task-oriented authoring tool integrated with ChatGPT for producing personalized educational content. Design considerations using the SDLC Waterfall Model and prompt engineering were discussed. The research involved a total of 25 participants: experts (n=5) and novice (n=20), who utilized the authoring tool to generate academic content. A 5-likert questionnaire that consisted of 41 items was designed to investigate the users’ agreement about the tool’s effectiveness, cognitive load, usability, and AI-associated challenges, with mean comparison and t-tests being used for analysis. The primary findings revealed overall positive impressions among users, particularly concerning the tool's efficiency and cognitive load management. Nevertheless, small differences in usability perceptions arose between experts and novices. These findings provide valuable insights for refining and augmenting AI-integrated authoring tools to better accommodate varying user requirements in the educational domain.

Applications of Rule-based Systems in Dental Decision Making: Scoping Review

Sara Aloufi, Mayada Alrige, Dalea Bukhary — Sun, 11 Aug 2024 00:00:00 +0000

This scoping review aims to explore and summarize the application of rule-based systems (RBSs) widely employed
in dentistry and to evaluate their performance and practical significance. We conducted a scoping review following the
methodology of PRISMA Extension for Scoping Reviews (PRISMA-ScR) on five databases: Web of Science, Scopus, Google
Scholar, Saudi Digital Library, and the IEEE Xplore. We searched for literature published in English up to October 2021. Two
reviewers evaluated each potentially relevant study for inclusion/exclusion criteria, and any discrepancies were resolved by a third researcher. Out of 303 searched studies, 19 fulfilled this review’s inclusion criteria. We identified two domains based on the methodology used in the included studies: (i) uncertainty management approaches employed in the RBS (n = 16) and (ii)
integrating machine learning techniques with the RBS (n = 5). The vast majority of included publications used fuzzy logic to
manage uncertainty (n = 11). A hybrid fuzzy RBS and neural network achieved the highest accuracy of 96%. The review also
found that periodontology was the most commonly addressed specialty. In an analysis of the current literature, RBSs were found reliable in assisting dental practitioners in decision-making. Clinical decision-making involves a high level of uncertainty, which explains the tendency to use fuzzy logic in RBSs. These systems can also be used as educational tools primarily for both
undergraduate dental students and less experienced dentists (e.g., dental interns, postgraduate, and junior dentists) to aid in making reliable decisions.

Context-Aware Recommendation System Using Matrix Factorization

Masarra Aldramley, Dimah Alahamdi — Sun, 11 Aug 2024 00:00:00 +0000

Abstract—In a commercial field, millions of new items would be added to the daily sales field. Suggesting proposed items to users for purchasing process is a critical point. Finding the best suggestions based on the user needs and behavior increase the sales productivity. Incorporating context information in recommendations process have been accompanied by many domains and applications. Different methods and strategies have been used to find recommendations. While time is an important factor for continuously updates and changes in the user preferences, incorporating it has been proved its effectiveness to enhance recommending performance. Time-aware recommender systems (TARS) has been used in a wide range of recommendation modeling. In this proposed paper, we focus to deal with three different context-aware algorithms. First, traditional matrix-factorization using explicit ratings. Second, enhanced version after dealing with time-target as basic factors for getting the results. Last, depend on the previous version, we enhance it by shrinking weights using the mathematical decay-function algorithm to improve prediction accuracy. we build our solutions and implement them using a real dataset of commercial website as our empirical case study. From our analysis and experiment, we finally evaluate the proposed model using different metrics on measuring relative performance of enhanced TARS over traditional MF.

Employing Sequence to Sequence Neural Network Model for XSS Attack Detection

Mohammad Alzahrani — Sun, 11 Aug 2024 00:00:00 +0000

Cross-site scripting (XSS) attacks are considered one of the most prevalent types of attacks and have caused huge damage to individuals and organizations in the form of economic loss and intrusion into privacy. Several detection techniques have been used to find known threats using signatures obtained from network traffic. Researchers have developed many techniques based on machine learning to identify attacks without depending on known signatures of already known attacks. While a number of neural network-based methods to detect XSS attacks have been proposed by security experts, no one has attempted to detect XSS attacks using a sequence neural network model. We have proposed a novel approach called a sequence-to-sequence neural network (seq2seq) model to detect cross-site scripting attacks without depending on signatures of known attacks. Using seq2seq model for XSS detection is based on extracting features from web application code segments, then using them to predict whether a script contains malicious code. The seq2seq model is represented as a two-layer neural network, with the first layer processing the training samples in sequential order and the second layer responsible for the classification of each data sample. This dataset consists of 10100 instances of malicious and benign JavaScript. The Pearson correlation method was used for feature selection. All the experiments were conducted using Tenser flow and Keras. The experimental results that proposed seq2seq achieved an accuracy of 99.8%.

Offline Signature Verification Using Deep learning and Genetic Algorithm

Abdulbaset Musleh, Abdoulwase Mohammed Obaid — Sun, 11 Aug 2024 00:00:00 +0000

The process of verifying signatures has wide-ranging applications in computer systems, including financial operations,
electronic document signing, and user identity verification. This approach has the advantage of community acceptance and
presents a less intrusive alternative than other biological authentication methods. Deep learning (DL) and Convolutional Neural
Networks (CNNs) have emerged as prominent tools in the field of signature verification, significantly enhancing the accuracy and effectiveness of these systems by effectively extracting discriminative features from signature images. However, optimizing the hyperparameters in CNN models remains a challenging task, as it directly affects the efficiency and accuracy of the models.
Currently, the design of CNN architectures relies heavily on manual adjustments, which can be time-consuming and may
not yield optimal results. To address this issue, the proposed method focuses on employing a genetic algorithm to evolve
a population of CNN models, thereby enabling the automatic discovery of the most suitable architecture for offline signature
verification. By leveraging the optimization capabilities of the genetic algorithm, the proposed approach aims to improve the
overall performance and effectiveness of the signature verification model. The effectiveness of the proposed method was evaluated using multiple datasets, including BHSig260-Bengali, BHSig260-Hindiin, GPDS, and CEDAR. Through rigorous testing, the approach achieved remarkable discrimination rates with a False Rejection Rate (FRR) of 2.5%, a False Acceptance Rate (FAR) of 3.2%, an Equal Error Rate (EER) of 2.35%, and an accuracy rate of 97.73%.

A Survey on the Integration of 6G, IoE, and Quantum Computing Technologies

Ahmed Alshaflut — Sun, 11 Aug 2024 00:00:00 +0000

This survey paper comprehensively examines the emerging integration of three pivotal technologies: 6G (Sixth-generation wireless systems), IoE (Internet of Everything), and Quantum computing. The combination of these cutting-edge technologies will be seen leading in the computer-based communications era. The paper starts by providing the underlying reasons for such an integration, which are driven by high demands for higher data speeds, connectivity, and massive computation powers. The main part of this paper deals with examining the personal traits of every kind and contemporary trends in this field. 6G is expected to be way ahead of 5G when it comes to throughput, with applications such as holographic communications and virtual reality to span internationally. IoE is an extension of the concept known as the Internet of Things (IoT), forming a broader inter-network where people, processes, data, and smart things are connected to turn raw data into useful, relevant information that can be acted upon. The emerging data technology of Quantum Computing offers an entirely different paradigm in terms of speedy and complicated computation and solving problems. This further leads to an exploration of the possible advantages and usefulness of combining these technologies. It is expected that this integration will give a big push towards the improvement of communication, information analysis, and computation power across sectors such as health, smart city and cyberspace. The paper also discusses some problems arising from such integration, like complexities on the technical end, security issues and major infrastructure needs. This paper provides basic knowledge for researchers, practitioners, and policymakers interested in technological integration by providing ideas on how 6G, IoE, and Quantum Computing can play major roles in defining future technology frameworks.

Examining the Impact of Personality on the Efficiency of Recommendation Systems

Asaeyl Alahmadi, Roaa Aldhahri — Sun, 11 Aug 2024 00:00:00 +0000

Recent research shows great promise for predicting personality from social media data. This preliminary small sample review provides an idea on the possibilities that social media data and social media platforms could provide for measuring personal- ity. The review suggests that scientifically designed online environments or applications could provide interesting possibilities for collecting and analysing personal, social and mass-behavioural data. Furthermore, social media users interest to self-present align well with the interests of personality researchers, which suggests valuable motivational resources. A theoretical framework of these possibilities is provided as well as experi- ences regarding the small sample review method used in this study. This paper discusses personality-aware recommendation systems. With the evolution of artificial intelligence (AI) nowadays, personality-aware recommendation systems are considered a new re- search field related to AI and the psychology of personality. Also, it solves the most common problems, which are cold start and sparsity of data, of the traditional recom- mendation systems. The results of our comprehensive search, address the core research questions related to efficiency, personality theory, and techniques.

DeepSTEMI: Artificial Intelligent Support System for Rapid Diagnosis and Treatment of ST-segment elevation Myocardial Infarction in Pre-hospital Emergency Medical Services at SRCA In Makkah Al-Mukarramah

Sun, 11 Aug 2024 00:00:00 +0000

This research addresses the critical imperative for improved cardiac care in pre-hospital emergency services through the integration of an artificial intelligence (AI) based diagnostic system. A survey involving 237 participants in Saudi Arabia illuminates the essential need to minimize on-site duration during cardiac emergencies, with unanimous agreement among participants regarding the pivotal role of AI in expediting responses, also demographic analysis provides valuable insights into participant trends, contributing to a comprehensive background understanding. The study proposes a methodological pipeline that encompasses key elements, including data augmentation, ResNet50 model training, and the development of a user-friendly AI assistant named DeepSTEMI. This AI assistant is designed to predict specifically ST-segment elevation myocardial infarction (STEMI) from given images and respond to initial treatment. Demonstrating robust binary classification performance, the ResNet50 model consistently exhibits high precision, recall, F1-score, and accuracy. A validated area under the curve (AUC) score of 0.98 underscores the model's discriminative prowess in distinguishing STEMI from normal cases. Emphasizing practical strategies, the study advocates for collaboration with the Saudi Red Crescent Authority, continuous model refinement, and system expansion to address a broader spectrum of cardiac conditions. Furthermore, the research highlights the importance of integrating real-time data feeds and incorporating continuous learning as pivotal elements to enhance diagnostic precision.