		FeedBack \| Contact Us \| Links \| Site Map

Welcome to Journal of Theoretical and Applied Information Technology

Submit Paper / Call for Papers

Journal receives papers in continuous flow and we will consider articles from a wide range of Information Technology disciplines encompassing the most basic research to the most innovative technologies. Please submit your papers electronically to our submission system at http://jatit.org/submit_paper.php in an MSWord, Pdf or compatible format so that they may be evaluated for publication in the upcoming issue. This journal uses a blinded review process; please remember to include all your personal identifiable information in the manuscript before submitting it for review, we will edit the necessary information at our side. Submissions to JATIT should be full research / review papers (properly indicated below main title).

Journal of Theoretical and Applied Information Technology
January 2024 | Vol. 102 No.1

Title:	LATENT MODELING FOR PREDICTING MULTIDIMENSIONAL DATA
Author:	YASMINA AL MAROUNI, YOUSSEF BENTALEB
Abstract:	The purpose of the current paper is to find and adapt a statistical method that tackles two main issues. First, prediction in a multidimensional context whether for quantitative or categorical data. Second, modeling a complex cause-effect relations. In particular, the use of structural modeling of latent and manifest variables to derive the regression equation. This leads us to the discussion of the two main methods within Structural Equation Modeling (SEM): Partial Least Squares Path Modeling (PLS-PM) and Linear Structural Relations (LISREL). Upon a thorough comparison of the two methods, it was determined that the best approach is PLS-PM. Nevertheless, it is essential to acknowledge that this method has its limitations. To address these shortcomings, the authors have proposed an adaptation of the PLS-PM. The paper concludes with the practical application of the developed method. This enabled us, using a small sample of non-quantitative variables, to model the phenomenon of cybercrime among children via latent and manifest variables and to predict whether a child might become a victim of cybercrime.
Keywords:	SEM, PLS, LISREL, PLS-PM, Path modeling, Categorical data, Missing values, Small sample size, Latent variables.
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	AN INTELLIGENT DEEP LEARNING BASED AQI PREDICTION MODEL WITH POOLED FEATURES
Author:	SANTHANA LAKSHMI V, AND VIJAYA M S
Abstract:	Airborne pollution poses a significant threat to public health, leading to detrimental health effects. Despite global economic growth, ensuring access to clean air has become increasingly challenging worldwide. The contamination of air occurs as dust particles and smoke, released by vehicles and industries, suspend into the atmosphere, exacerbating the challenge of providing clean air for people. Hence, it is imperative to predict the Air Quality Index (AQI) to safeguard the lives of people, especially considering the severe health effects caused by the inhalation of small particles. This paper outlines a deep learning methodology for constructing Air Quality Index (AQI) prediction models. The models utilize hourly meteorological data and pollutant information, aiming to fulfill the critical requirement for precise assessments of air quality. The aim of this paper is to formulate predictive models for AQI in Thiruvananthapuram, Kerala, employing deep learning algorithms, thereby addressing the escalating challenge of air pollution in the region. Deep neural network architectures, such as Long Short-Term Memory (LSTM), Bidirectional Long Short-Term Memory (BI- LSTM), and Gated Recurrent Unit (GRU), are implemented to construct the prediction model. When compared to other algorithms, GRU demonstrated promising outcomes. The findings of this research contribute not only to the advancement of AQI prediction models but also highlight the practical significance of employing deep learning techniques for accurate and timely air quality assessments. The outcomes have practical implications for public health and environmental management, providing a basis for informed decision-making in mitigating the adverse effects of air pollution.
Keywords:	Ammonia, CO, Pollution, Prediction Models, Meteorological Data
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	DIMENSIONS OF ENTREPRENEURIAL ORIENTATION AND ITS IMPACT ON BUSINESS AND SOCIAL PERFORMANCE OF STATE-OWNED ENTERPRISES
Author:	SHARUL NIZAL SHARIPPUDIN, NOMAHAZA MAHADI, WAN NORMEZA WAN ZAKARIA
Abstract:	State-owned enterprises (SOEs) are critical for nations’ socio-economic development. SOEs carry two main roles which are to achieve business goals to satisfy the interest of shareholders and at the same time to pursue social goals that create public values. However, in the current increasingly challenging business environment, SOEs face difficulties to pursue both goals simultaneously. Thus, based on resource-based view (RBV), this study examined the role of entrepreneurial orientation (EO) as a dynamic capability derived from innovativeness, proactiveness and risk-taking in influencing the achievement of business and social performance. This study applied quantitative method and data was collected from among the leadership groups of government linked companies (GLCs) in Malaysia. The data was analyzed using the Partial Least Square-Structural Equation Model (PLS-SEM) technique. The analysis revealed three key findings as following, (1) innovativeness has positive relationship with business and social performance; (2) proactiveness has positive relationship with business performance but has negative relationship with social performance; (3) risk taking has negative relationship with both performances. Therefore, by examining EO as multidimensional construct and its impact on business and social performance, this study provided insight of which dimensions of EO is significant in value creation process. This study also provided useful insight to the policy makers and management the best way to improve SOEs’ performance in meeting business and social goals.
Keywords:	Entrepreneurial Orientation, State-Owned Enterprise, Resource-Based View, Business Performance, Social Performance
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	INTERACTIVE USER INTERFACES IN THE DIGITAL WORLD MAKE THE APPLICATION ATTRACTIVE AND EASIER FOR USER ACCESS
Author:	WAHYU SARDJONO, WIDHILAGA GIA PERDANA
Abstract:	In today's digital era, Indonesian people rely heavily on electronic media as a tool to support life. The digital era, accompanied by the development of advanced technology, has proven to have a good impact on the people of Indonesia. When talking about electronic media, one cannot skip discussing the digital world. The digital world is a representation of cyberspace that can make it easier for Indonesian people to access applications or websites. Intermediary tools used in accessing the digital world are electronic media, such as mobile phones, laptops, and so on. In an application, of course, there are various components in it, for example, such as the User Interface (UI). User Interface (UI) has functions to connect users with the system. The ease and effectiveness of the User Interface (UI) in an application/website have a very large impact on user interest in using the application/website, where this interest can encourage user confidence to always use it. This has been proven by 45.1% of 110 application users who strongly agree with the statement that the efficiency of a User Interface (UI) greatly influences user comfort in carrying out an activity with the system. This study uses literature study, observation, and quantitative research methods. The existence of an interactive User Interface (UI) in the digital world makes the application/website look attractive and makes it easier for users to access.
Keywords:	Application Digital, Electronics, User Interface, Website, Interactive User
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	THE USE OF CONVOLUTIONAL NEURAL NETWORK WITH EFFICIENTNET-B0 ARCHITECTURE IN BRAIN TUMOR CLASSIFICATION USING FLASK
Author:	VINSENSIUS HARYO BHAKORO HADI, ACHMAD BENNY MUTIARA, RINA REFIANTI
Abstract:	Brain tumors pose a significant health challenge, often evading early detection and resulting in fatal outcomes. Despite the assistance of imaging technologies like CT-Scan and MRI, achieving timely and accurate diagnoses remains a formidable task. This study advocates for the implementation of Convolutional Neural Networks (CNNs) to facilitate rapid and precise brain tumor detection. Specifically, we assess the performance of EfficientNet-B0, an advanced CNN architecture, in comparison to other CNN architectures for classifying brain tumors in MRI images. Our dataset comprises 3264 images across glioma, normal, pituitary, and meningioma classes. Testing involved various scenarios for epochs and optimizers, including Adam and RMSProp. Model testing results, analyzed through a confusion matrix, revealed an impressive average precision, recall, and F1-Score, all reaching 98%. The best-designed model accurately predicted glioma, normal, pituitary, and meningioma tumor types. Furthermore, the successful implementation of the classification model into a website using Python and the Flask framework signifies its potential for practical applications, enhancing accessibility and usability.
Keywords:	Convolutional Neural Network, EfficienNet-B0, Brain Tumor, Flask, Python.
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	ALGORITHMIC COMPOSITION USING GATED RECURRENT UNIT FOR NATIONALISTIC MUSIC
Author:	KHAFIIZH HASTUTI , ERWIN YUDI HIDAYAT
Abstract:	The study scrutinizes the implementation of the Gated Recurrent Unit within Recurrent Neural Networks for constructing nationalistic music. The GRU model demonstrates the capability to algorithmically emulate patriotic melodies from original compositions, thereby highlighting the transformational role of machine learning in crafting intricate musical structures. The effectiveness of the GRU model is further evaluated through the Turing Test, revealing a significant 46.5% misidentification rate. This evidence underlines the model's success in producing complex compositions that bear a striking resemblance to human-created pieces. Ultimately, these findings contribute to the broader understanding of GRU's potential in innovative music composition, thereby facilitating the enhancement of nationalism through the potent medium of music.
Keywords:	Algorithmic Composition, GRU, Nationalistic Music, Turing Test, LSTM
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	CONTINUOUS AND MUTUAL LIGHTWEIGHT AUTHENTICATION FOR ZERO-TRUST ARCHITECTURE-BASED SECURITY FRAMEWORK IN CLOUD-EDGE COMPUTING-BASED HEALTHCARE 4.0
Author:	WALEED ALMUSEELEM
Abstract:	Healthcare 4.0 is a heterogeneous environment in which many smart medical devices are connected to provide timely healthcare services. As the next generation of Healthcare 4.0 enables more digitized and interconnected services across multiple devices and communication technologies, the possibility of potential attack also expands significantly. Critical healthcare deals with highly sensitive patient data and has to fulfill strict regulatory requirements. Thus, incorporating Zero Trust Architecture (ZTA) is paramount to offer a robust framework that ensures safety and security against evolving threats. This work proposes a framework that exploits ZTA based continuous lightweight mutual authentication strategy for Healthcare 4.0 to accomplish secure data transmissions among the devices, edges, and cloud server. It is a flexible and lightweight authentication strategy that considers all the entities in Healthcare 4.0 untrusted and enables continuous authentication during every session to ensure high security against various vulnerabilities. The continuous and mutual authentication based is accomplished on two different levels. Firstly, the dynamic Hash-based Message Authentication Code (HMAC) based continuous mutual lightweight authentication is exploited two different transmissions that are Device-to-Device (D2D) and Device-to-Edge (D2E). Thus, it attains a better tradeoff between security and resource consumption over resource-constrained healthcare 4.0 devicesSecondly, the framework employs the Elliptic Curve Cryptography-Advanced Encryption Standard (ECC-AES) based heavyweight authentication and Identity Based Access Control (IBAC) to enable secure authorization and access control in Edge to Cloud Server (E2C) transmission. Further, the framework analyzes its efficiency in three ways: Scyther-tools-based security analysis, theoretical analysis, and simulation-based analysis. Moreover, the Contiki/Cooja-based simulations proved that the proposed framework is a strong competitor among various D2D and D2E authentication protocols in healthcare 4.0 environments.
Keywords:	Zero-trust architecture, Device-to-device, Device-to-Edge, Edge-to-cloud, Authentication, Authorization and Access Control
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	REVOLUTIONIZING EMERGENCY RESPONSE: 5G AND IOT ENABLED ADR SYSTEM FOR UAV DELIVERY OF AID IN A CATASTROPHE
Author:	BHARATHI S, P. DURGADEVI
Abstract:	Intelligent transportation systems (ITSs) are being enabled by the Internet of Things (IoT) and 5G. ITSs have the potential to enhance roadway security in smart cities. As a result, ITSs are gaining traction in both the industry and academia. Vehicle numbers are increasing due to the fast growth in population. As a result, the frequency of traffic accidents is growing. The majority of the moment, accidents aren't identified or communicated to nearby hospitals and family in an appropriate manner. This absence of prompt medical attention and first assistance. In a matter of minutes, you might lose your life. To handle all of these issues, an intelligent system is required. Although numerous ICT-based solutions for identifying accidents and rescue efforts have been offered, these systems are not interoperable with every vehicle and are also expensive. As a result, we presented a smart city accident detection and reporting system (ADR) that is less costly and compatible with any vehicle. Our plan strives to enhance the quality of transportation while keeping costs reasonable. In this context, we created a simulation that gathers information from the sensors about pressure, acceleration, captures moment, location of the occurrence, and force of the accident. The measure of speed contributes to the accuracy of accident recognition. The information gathered is then analysed for accident identification. A GPS system should also alert relatives, the police, and the closest medical center. The hospital sends a UAV (drone) along with a first aid kit and a motor ambulance to accident location. The road transport and roads accident repository's real dataset are used to produce simulation results. The proposed plan shows promise when measured against current approaches in terms of accuracy and response time.
Keywords:	Accident Detection, Accident Prevention, 5G, IoT, ADR
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	ENHANCING ENGLISH LEARNING OUTCOME PREDICTIONS: A HYBRID APPROACH INTEGRATING GRADIENT BOOSTING AND K-NEAREST NEIGHBOURS TECHNIQUES
Author:	MYAGMARSUREN OROSOO, KATHARI SANTOSH, LATEFA ALFRYAN, DR. SAFEER PASHA M, DR GRANDHI PRASUNA, MANIKANDAN RENGARAJAN
Abstract:	In the realm of educational data analysis, accurately predicting English learning outcomes holds paramount significance. This study introduces an innovative approach by proposing an ensemble model that synergistically combines two powerful machine learning techniques: Gradient Boosting and K-nearest Neighbours. Through comprehensive data pre-processing and feature engineering, the proposed ensemble model harnesses the strengths of both algorithms. Gradient Boosting excels in capturing intricate patterns and dependencies within the data, while K-nearest Neighbours excels in uncovering local relationships and proximity-based insights. The ensemble model strategically amalgamates the predictive insights from both methodologies, capitalizing on their complementary nature. By leveraging this hybridized approach, the ensemble model endeavours to provide enhanced accuracy and robustness in predicting English learning outcomes. The effectiveness of the proposed ensemble model is rigorously evaluated through comprehensive experimentation and performance assessments, demonstrating its potential to offer an advanced and holistic solution for predicting English learning outcomes with practical implications for educational institutions and stakeholders. Our ensemble technique obtains a remarkable accuracy percentage of 99.5% through thorough examination. This result shows the effectiveness of ensemble techniques in educational predictive analytics and draws attention to their potential to fundamentally alter educational decision-making procedures. This research stimulates improvements in pedagogical practices and eventually contributes to the enrichment of the learning experience by providing educators, administrators, and policymakers with a trustworthy technical instrument for forecasting English learning results.
Keywords:	Gradient Boosting, K-nearest Neighbours, English Learning Outcomes.
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	EXPLORING THE DYNAMICS OF EDUCATIONAL FEEDBACK NETWORKS WITH GRAPH THEORY AND LSTM-BASED MODELING FOR ENHANCED LEARNING ANALYTICS AND FEEDBACK MECHANISMS
Author:	ASFAR H. SIDDIQUI, KATHARI SANTOSH, DR. MOHAMMED SALEH AL ANSARI, BADUGU SURESH, MRS. V. SATHIYA, PROF. TS. DR. YOUSEF A. BAKER EL-EBIARY
Abstract:	The promise of learning analytics to transform education by offering insightful data on student learning patterns and enabling personalized feedback mechanisms has attracted a lot of interest in recent years. In order to improve learning analytics and feedback mechanisms, this study uses graph theory and Long Short-Term Memory (LSTM) based modelling to analyze the dynamics of educational feedback networks. The study makes use of a sizable dataset made up of educational interactions from various learning settings, including student contributions, evaluations, and teacher comments. A network model of these interactions is created using graph theory, where nodes stand in for students, teachers, and educational materials, and edges for feedback linkages. This network-based strategy makes it possible to see the educational environment as a whole and makes it easier to analyze feedback dynamics. Furthermore, taking into account the sequential character of educational encounters, LSTM-based models are created to represent temporal relationships within the feedback networks. These models allow for the evaluation of feedback quality, the identification of influential nodes, and the prediction of future feedback patterns. A thorough foundation for comprehending the complex dynamics of educational feedback networks is provided by the combination of graph theory with LSTM-based modelling. This paper gives a distinctive viewpoint on the evaluation of educational feedback networks by fusing graph theory with LSTM-based modelling. The suggested framework has the power to improve educational practices, guide selections for instructional design, and encourage student achievement.
Keywords:	Learning analytics, graph theory, LSTM
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	A NEW METHOD FOR GETTING THE OPTIMAL NUMBER OF CLUSTERS BY K-MEANS USING THE WEIGHTED BARYCENTER
Author:	JEDDIN SARA, BENTALEB YOUSSEF
Abstract:	Clustering is a popular unsupervised algorithm in data science used to group similar data points together. One of the major challenges of using clustering algorithms is to determine the optimal number of clusters. To achieve this step, the Elbow method is a commonly used technique to identify the optimal number of clusters and often used in conjunction with K-means algorithm. However this method has some limitations and disadvantages, it is based on minimizing the sum of squared distances between each data point and the centroid of its assigned cluster that’s why it provides information about the homogeneity inside clusters but it can’t provide information about how is the distance between clusters. This paper suggests an enhanced Elbow algorithm that utilizes the concept of weighted barycenter to address the issue of group separation. The improved method is based on calculating the distance between the barycenter of the clusters identified by the K-means.
Keywords:	K-means, Elbow, Barycenter.
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	GRAPH CONVOLUTIONAL NEURAL NETWORK FOR IC50 PREDICTION MODEL WITH DRUG SMILES GRAPHS AND GENE EXPRESSIONS OF AMYOTROPHIC LATERAL SCLEROSIS
Author:	DEVIPRIYA S, VIJAYA M S
Abstract:	IC50 prediction for neurodegenerative disorders like Amyotrophic Lateral Sclerosis is crucial in biomedical studies. Traditional machine learning models that use molecular descriptors and gene expression for building IC50 prediction models produce less accuracy and also most of the descriptors created by different tools are irrelevant and undefined. In this paper, a Graph Convolutional Neural Network, a deep learning algorithm, is employed for constructing a more precise IC50 prediction model. The model leverages the structural properties of drug molecules represented in graph format, and incorporates gene expression data as global features. So, the model is able to learn drug-gene interactions better. The drug-gene interactivity is learned by the model without drug-induced gene expressions as it is not found for most of the diseases. The work is implemented with well-known and most relevant 80 drugs related to ALS based on the pIC50 values of 32 protein targets of ALS disorder. The Canonical Smiles graph and their corresponding IC50 values of 80 drugs have been derived from the ChEMBL databases. Based on information from the Repurposing Hub in the Depmap database gene expression data for drug-related genes connected with ALS-related conditions is collected. The predictive results show that the proposed GCNN model with fine-tuned hyperparameters achieves MAE of 0.18, RMSE of 0.16 and R2 Score of 0.90.
Keywords:	IC50, Gene Expression, Graph Convolutional Neural Network, SMILES, Prediction
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	CYBERSECURITY: MALWARE MULTI-ATTACK DETECTOR ON ANDROID-BASED DEVICES USING DEEP LEARNING METHODS
Author:	MUSTAFA ABABNEH, AAYAT ALJARRAH
Abstract:	Android-based devices are currently a prime target for cyber-attackers. New malware is being developed and released, with devastating effects on sensitive information lost and ransom payments. Android developers and users continue to look for holistic methods of detecting all types of malware instead of individual ones. The aim of this study is to test the combined impact of deep learning (DL) methods on detecting malware with multi-attack features on Android devices. A malware multi-attack detector (MMAD) combined DL methods: deep neural networks (DNN), recurrent neural networks (RNN), convolutional neural networks (CNN), multilayer perceptron’s (MLP), and end-to-end (E2E). Each of these methods detects specific types of malware. Different types of malware attacks, including benign ones, were used to train and test the MMAD model. Experimental results indicated that the proposed MMAD model was efficient in detecting eleven types of malware attacks with a high and constant multi-classification capability. Our results with 96.54% accuracy, 95.38% precision, 92.65% recall, and a 94.66 F-score showed that the MMAD approach is effective, efficient, and simple to use.
Keywords:	Android Devices; Malware Multi-Attack Detectors; Deep Learning Methods.
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	DIGITAL HEALTH SYSTEM AND E-HEALTH IN THE FOLLOWING COUNTRIES: UNITED KINGDOM, NORWAY, SWEDEN, DENMARK, GERMANY AND UNITED STATES
Author:	HILALI OUMAIMA, SOULHI AZIZ
Abstract:	The Covid19 pandemic, the rise of chronic diseases and the high rate of old age worldwide are factors that have impacted the world's healthcare systems. This has led to hospital congestion, exhausted medical practitioner and a strained healthcare system. New information technologies, digitization and artificial intelligence are the levers needed to restore a resilient, robust healthcare system that can face up to the uncertainties of our current world. In this article, we present health systems in several countries to know the architecture, strengths, and weaknesses of the different systems. We also present the state of progress of the digitization of healthcare systems in these countries, the contribution of digitization to the quality of healthcare, and the obstacles and problems encountered when digitizing the system.
Keywords:	Digital Health System, Digital Health Policy, Data Health, Telemedicine, E-Health, Electronic Health Record
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	A PRELIMINARY QUANTITATIVE STUDY ON THE PARENTAL ROLE IN PROTECTING CHILDREN FROM ELECTRONIC BLACKMAIL: THE CASE OF DUBAI/U.A.E.
Author:	NABIH ABDELMAJID , DIMITRIOS XANTHIDIS , SUJNI PAUL
Abstract:	The swift integration of information technologies has resulted in the pervasive utilization of social networking sites, with students emerging as some of the most frequent users. Consequently, this increased engagement exposes students to potential cyber intrusions. This preliminary quantitative study aims to investigate the role of parents in safeguarding their children against electronic hackers and guiding them in the effective and secure use of information technologies. The research involved conducting a survey that considered various demographic factors, with a particular focus on the educational level of parents. The findings underscore the crucial need for parents to take an active role in protecting their children, even when the children themselves are proficient in the use of these technologies.
Keywords:	Parental Control, Online Threats, Security, Privacy, Social Media.
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	IMPROVING PERFORMANCE, CRYPTOGRAPHIC STRENGTH OF THE POST-QUANTUM ALGORITHM NTRUENCRYPT AND ITS RESISTANCE TO CHOSEN-CIPHERTEXT ATTACKS
Author:	ELENA REVYAKINA, LARISA CHERCKESOVA, OLGA SAFARYAN, NIKITA LYASHENKO
Abstract:	This work is devoted to the development of modification of the post-quantum NTRUEncrypt algorithm to improve its performance and resistance to modern cyberattacks which in turn allows it to be used in modern practical applications. To accomplish this goal, the authors take an approach that involves optimization of polynomial multiplication, which is the most computationally complex part of the algorithm. The Anatoly Karatsuba’s algorithm has been successfully applied to significantly increase the speed of key generation and encryption. In spite of the fact that this algorithm involves recursion, it allows to improve performance of the NTRUEncrypt algorithm, since it allows to speed up polynomial multiplication. Another notable improvement is developing the countermeasure against chosen ciphertext cyberattack. Resistance against this type of cyberattacks can be accomplished by employing cryptographic hash function to reject messages sent by malicious users. Performance test is carried out to estimate the average time required to generate keys and perform both encryption and decryption of a message. From the results of the performance test, it has been concluded that in the implemented modification of the algorithm, key generation, encryption and decryption require less time compared to the classical algorithm. The most significant performance gain has been achieved for the key generation stage since it involves numerous complex computations that can be performed much faster due to utilizing the Karatsuba’s algorithm. Based on the modified algorithm, an asymmetric encryption system with a graphical interface has been implemented, which allows users to transfer messages with ensured protection against all modern attacks, including quantum cyberattacks.
Keywords:	Post-Quantum Algorithm, NTRUEncrypt Cryptosystem, Cyberattack, Polynomial Multiplication, Karatsuba Algorithm.
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	AN ADAPTIVE PRIVACY PRESERVING BASED ENSEMBLE LEARNING FRAMEWORK FOR LARGE DIMENSIONAL DATASETS
Author:	CH. NANDA KRISHNA , K.F. BHARATI
Abstract:	With the rapid expansion of data, increasing computational power, and the complexity of high-dimensional datasets, it is of utmost importance to integrate a novel privacy-preserving model into deep learning frameworks. These frameworks, commonly utilized in traditional machine learning applications, heavily rely on extensive databases. Consequently, safeguarding the sensitive patterns generated by these approaches before they are uploaded to cloud storage becomes imperative. However, the development and implementation of a privacy-preserving deep learning model, specifically designed for highly dimensional cloud data, pose significant challenges. In cloud computing, traditional privacy-preserving deep learning frameworks prioritize data transformation methodologies over cryptographic approaches due to the substantial computational memory and time requirements involved. This preference is crucial to ensure privacy while simultaneously distributing multiple datasets in real-time multi-user applications. As the size of these applications expands, traditional privacy-preserving deep learning models require substantial computing resources to effectively preserve the intricate patterns generated by machine learning algorithms. To overcome these challenges, a unique privacy-preserving deep learning model is constructed, leveraging high-dimensional datasets and employing data partitioning techniques. Experimental findings validate the exceptional computational accuracy achieved by this model while simultaneously preserving privacy in the patterns, surpassing the performance of existing models.
Keywords:	Privacy Preserving, Machine Learning, Cryptography.
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	HEURISTIC EVALUATION OF LEARNING TECHNOLOGY FOR SPECIFIC DISABLED LEARNER
Author:	SYAZWANI RAMLI , HAZURA MOHAMED , ZURINA MUDA
Abstract:	A disabled learner is a special needs child who requires full attention and a specific approach, particularly one who necessitates learning in special education. These learners need a unique approach to tackle their limitations since they have lower cognitive skills than a normal learner. In the current 21st century, many types of learning technologies can assist them in the learning process. However, not all learning technologies are suited to the capacity of disabled learners. The paper aims to discuss the results of the heuristic evaluation that has been carried out using a 3M learning application among disabled learners in the classroom setting. The heuristic evaluation involved six experts in determining the flaws in designing the 3M learning application as an educational technology for disabled learners. Based on the heuristic evaluation that has been conducted, the results achieved exceeded 77.5% of the usability percentage. Hence, it shows that the 3M learning application has a good user experience level that can be used by a disabled learner to maximize and increase understanding during the learning process.
Keywords:	Disabled Learner, Educational Technology, Heuristic Evaluation, Special Education
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	ELEVATING SENTIMENT ANALYSIS WITH RESILIENT GREY WOLF OPTIMIZATION-BASED GAUSSIAN-ENHANCED QUANTUM DEEP NEURAL NETWORKS IN ONLINE SHOPPING
Author:	G.M. BALAJI , K. VADIVAZHAGAN
Abstract:	The rise of online shopping reflects a significant change in consumer behavior, with more people drawn to digital marketplaces due to unparalleled convenience, extensive product variety, and competitive pricing offered by online platforms. Product reviews have become a cornerstone within this digital retail landscape, offering invaluable guidance to customers and business proprietors. Shoppers rely on reviews to make well-informed decisions, gaining insights into product quality and functionality, while entrepreneurs utilize this feedback to refine their offerings and elevate customer satisfaction. The analysis of sentiments embedded within product reviews presents formidable challenges due to the intricacies of human language and the sheer volume of data. To address this tough challenge, this paper introduces Resilient Grey Wolf Optimization-based Gaussian-Enhanced Quantum Deep Neural Networks (RGWO-GEQDNN). This novel approach amalgamates the robust, Resilient Grey Wolf Optimization with Gaussian-enhanced Quantum Deep Neural Networks, providing a potent solution for efficient and accurate sentiment analysis within product reviews. RGWO-GEQDNN emphasizes the innovative fusion of nature-inspired optimization and quantum computing principles, promising a breakthrough in sentiment analysis. To assess the performance of RGWO-GEQDNN against state-of-the-art algorithms, Amazon product review dataset is utilized. The results underscore the superiority of RGWO-GEQDNN in accurately classifying sentiment from product reviews, highlighting its transformative potential in the e-commerce landscape.
Keywords:	Sentiment, Reviews, Online Shopping, Classification, Amazon, Neural Network
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	SYSTEM IDENTIFICATION AND FUZZY CONTROLLER DESIGN OF TETHERED UNMANNED UNDERWATER VEHICLES (TUUV) USING DEEP REINFORCEMENT LEARNING CONTROLLER TO AVOID COLLISION
Author:	DEVJANI BHATTACHARYA , PUTTAMADAPPA C
Abstract:	In this paper, the main aim is to develop an effective method to avoid collisions underwater. Therefore, a novel method on fuzzy logic controller (FLC) (Fuzzy type 2) method and system identification have been designed with the standard vehicle depth and pitch control dynamics parameters, along with the equations of tethered Unmanned underwater vehicles (TUUV) are elaborately discussed in order to avoid movable obstacles found in underwater. Moreover, the fuzzy controllers help to accumulate appropriate information from the sonar system. Hence utilizing the obtained sonar data or information, the fuzzy controller defines the attack angle and speed using the movement captured through the underwater vehicle. The current research has used Deep Reinforcement Learning Controller for avoiding collision and normally helps to acquire knowledge in accordance with the success rate. The information gathered was utilized to provide an effective outcome in the process. The simulation results of the current study with the appropriate definition of variables and the performance evaluation will be estimated with the obtained output in terms of controlling parameters, including Depth for both positive and Negative, Depth rate, Pitch response, Trajectory, and Tracking for TUUV has been explained. The intelligent framework characteristics of the system identification and fuzzy controller design of TUUV utilizing a novel method - Deep Reinforcement Learning Controller to avoid collision has the capability to procure efficient and better results in avoiding movable obstacles found underwater. Eventually, the simulation results of the proposed novel technique allow the underwater vehicle to securely navigate by avoiding the obstacles in the desired path.
Keywords:	Tethered Unmanned Underwater Vehicles, Deep Reinforcement Learning, Fuzzy Logic Controller, Collision Avoidance, Pitch response, Trajectory, and Tracking
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	EMPOWERING ONLINE SHOPPING SENTIMENT ANALYSIS USING TENACIOUS ARTIFICIAL BEE COLONY INSPIRED TAYLOR SERIES-BASED GAUSSIAN MIXTURE MODEL (TABC-TSGMM)
Author:	G.M. BALAJI , K. VADIVAZHAGAN
Abstract:	Sentiment Analysis has become increasingly important in online shopping, where consumers rely on reviews and feedback to make informed purchasing decisions. However, accurate sentiment classification poses challenges, such as handling nuanced language and varying review lengths. This study introduces a novel approach called the Tenacious Artificial Bee Colony inspired Taylor Series-based Gaussian Mixture Model (TABC-TSGMM) to address these challenges.TABC-TSGMM leverages two key components: the Taylor Series-based Gaussian Mixture Model (TSGMM) and the Tenacious Artificial Bee Colony (TABC). TSGMM captures complex sentiment patterns in text data, while TABC optimizes the model’s performance through an intelligent search strategy.TABC enhances TSGMM by optimizing model parameters, making the sentiment analysis process more robust and accurate. To evaluate the effectiveness of TABC-TSGMM, we conducted experiments on an Amazon product review dataset, focusing on electronic products. The results demonstrated superior classification accuracy, showcasing the potential of this approach to empower online shopping sentiment analysis.
Keywords:	Sentiment Analysis, ABC, GMM, Taylor Series, Local Search. Optimization
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	OPTIMIZING DIGITAL EDUCATION: THE IMPACTS OF BIG DATA ON E-LEARNING IN THE HEALTH SECTOR
Author:	D. BENAMMI, S. BOUREKKADI, A. MASSIK, B. OUMOKHTAR
Abstract:	Online education, often known as e-learning, is rapidly emerging as a formidable substitute for traditional classroom instruction, thanks to the revolutionary shift in how people think about and engage with new forms of digital media traditional. The convenience of e-learning, which lets students access course materials whenever and whenever they choose, is a major factor in this paradigm change. But the capacity to tailor itself to the specific requirements of each student is the very heart of effective online education. Because every student is different and has their own preferred methods of learning as well as their own speed and degree of comprehension, tailoring lessons to each individual's needs is crucial. Big Data becomes a game-changer in this context. From users' activities and performance on the platform to their preferences in terms of content and the nature of their real-time interactions, e-learning platforms gather vast amounts of data. Individual learning patterns may be uncovered by mining this data, which is often varied and complicated. To interpret these trends and tailor the learning experience appropriately, Big Data provides a game-changing solution. Platforms may adapt their content, difficulty level, and even pedagogy to match the unique requirements of each student by evaluating data in real-time. Through the provision of a personalized experience that optimizes information retention, this real-time modification contributes to the development of a more engaging and effective learning environment.
Keywords:	E-Learning, Big Data, Personalization, Educational Optimization, Digital Transformation.
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	ENHANCING FAKE NEWS DETECTION ON SOCIAL MEDIA THROUGH ADVANCED MACHINE LEARNING AND USER PROFILE ANALYSIS
Author:	ZAHIR ABBAS KHAN , REKHA V
Abstract:	Social media news consumption is growing in popularity. Users find social media appealing because it's inexpensive, easy to use, and information spreads quickly. Social media does, however, also contribute to the spread of false information. The detection of fake news has gained more attention due to the negative effects it has on society. However, since fake news is created to seem like real news, the detection performance when relying solely on news contents is typically unsatisfactory. Therefore, a thorough understanding of the connection between fake news and social media user profiles is required. In order to detect fake news, this research paper investigates the use of machine learning techniques, covering important topics like feature integration, user profiles, and dataset analysis. To generate extensive feature sets, the study integrates User Profile Features (UPF), Linguistic Inquiry and Word Count (LIWC) features, and Rhetorical Structure Theory (RST) features. Principal Component Analysis (PCA) is used to reduce dimensionality and lessen the difficulties presented by high-dimensional datasets. The study entails a comprehensive assessment of multiple machine learning models using datasets from "Politifact" and "Gossipofact," which cover a range of data processing methods. The evaluation of the XGBoost classification model is further enhanced by the analysis of Receiver Operating Characteristic (ROC) curves. The results demonstrate the effectiveness of particular combinations of features and models, with XGBoost outperforming other models on the suggested unified feature set (ALL).
Keywords:	Linguistic Inquiry and Word Count features, Machine learning techniques, Principal Component Analysis (PCA), Rhetorical Structure Theory (RST) features, Unified feature set (ALL), User profile features (UPF)
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	DIGITALIZATION AND CLASSIFICATION OF SCANNED ECG USING CONVOLUTIONAL NEURAL NETWORK
Author:	AICHA CHOUMAD FALL, MOHAMED EL HACEN MOHAMED DYLA, MOUSTAPHA MOHAMED SALECK, NAGI OULD TALEB, TAOUFIQ GADI, MOHAMEDADE FAROUK NANNE
Abstract:	The electrocardiogram (ECG) is a non-invasive test that shows and records the electrical activity of the heart. However, the result is given on thermal graph paper which deteriorates rapidly with time. In this paper we work on two objectives, the first is the presentation of a new approach to convert scanned images to extract a good quality signal using computer vision methods. Each image is converted to a gray level, a region of interest is selected, then the bilateral filter method is applied, optimal for preserving the signal contour before binarization by Otsu’s method, then we complete with morphological operations. The second objective is the classification of the resulting images from the above method into two categories normal and abnormal. The experimental result of the images collected at the National Cardiology Center of Nouakchott shows the superiority of our approach over the global thresholding and Otsu methods. This ECG digitization approach is an accurate and reliable method for efficient storage and analysis. The results of the binary classification on our test base showed an accuracy of 97%.
Keywords:	Electrocardiogram (ECG), Digitization; Bilateral Filter, Otsu's Thresholding, CNN
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	INTRUSION DETECTION SYSTEMS IN INTERNET OF THINGS: A RECENT STATE OF THE ART
Author:	RACHID HDIDOU , MOHAMED EL ALAMI
Abstract:	Recently, the Internet of Things (IoT) has become a main technology in several areas such as smart networks, smart homes, smart cities, and others. By 2025, it is expected that more than 75 billion objects will be connected. This increase in internet-related objects implies the growth of cybercrimes against IoT networks. Since IoT is a set of heterogeneous objects, standard security techniques such as firewalls and antivirus are not sufficient to properly secure IoT infrastructures. This highlights the need to use flexible solutions such as Intrusion Detection Systems (IDS) which is the subject of our research. Our main objective is to determine the flaws and limitations of the existing solutions. To achieve this goal, we analyzed more than 60 articles on Intrusion Detection Systems in the Internet of Things. In this paper, we presented a taxonomy of Intrusion Detection Systems and a study on the architecture of the Internet of Things as well as attacks against the Internet of Things. Finally, we presented a detailed state of the art with the Problems, limitations of existing solutions, and open research issues for future research.
Keywords:	IDS, IoT, Intrusion Detection System, Internet of Things, IoT Security.
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	DEPTHWISE SEPARABLE CONVOLUTION RESNET MODEL FOR SENTIMENT ANALYSIS IN AMAZON E-COMMERCE WEBSITES
Author:	S J R K PADMINIVALLI V, M. V. P. CHANDRA SEKHARA RAO
Abstract:	In this era of modern world, exponential growth in e-commerce websites helps people buy necessary products online because of delivering to everyone’s doorstep. Consumers no need to get out of their home to buy products, through the websites they able to view wide variety of brands for each product. Since customers rely on e-commerce websites, the value of rating is important for business growth. To buy the products online, people solely rely on the review comments of products before purchasing the products. The reviews are sometimes lengthy, tedious and deceptive, in such situation sentiment analysis used which identify sentiment in reviews by investigating and extracting the views. This paper describes the depthwise separable convolution resnet model for sentiment analysis in amazon e-commerce website such that reviews are collected from publicly available dataset. On collecting the reviews, pre-processing of data by using Stemming, Lemmatization, Tokenization and Stop word removal for cleaning the data. In sentiment analysis, for reducing the dimensionality features are extracted to create a smaller set of features and feed this lower-dimensional data to classifiers to predict the sentiment polarity of the entire text. Thus review related features extracted by Latent Semantic Analysis (LSA) and Improved Term Frequency- Inverse Document Frequency (ITF-IDF) methods. From the review related features, aspect features are extracted by evaluating the weight matrices. Finally extracted features are fed to proposed model for the detection and classification of the consumers’ sentiment into positive, negative and neutral. The model is evaluated based on performance metrics and compared with existing techniques, proposed model obtains an accuracy of 98.2% which is higher than existing methods. Thus proposed approach helps manufacturers improve their products based on user feedback.
Keywords:	Latent Semantic Analysis, Term Frequency- Inverse Document Frequency, Stemming, Lemmatization, Tokenization, Stop Word Removal, Depthwise Separable Convolution Resnet Model.
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	MULTI-TASK LEARNING FOR MONKEYPOX SKIN LESION SEGMENTATION AND CLASSIFICATION USING CNN AND ROOTSIFT
Author:	KRISHNAN T , DR. SELVAKUMAR K , DR. VAIRACHILAI S
Abstract:	Introduction: The analysis of skin lesions is critical for diagnosing skin diseases accurately. In this research, we present an innovative approach that addresses the tasks of identifying and classifying skin lesions, with a focus on lesions caused by the monkeypox virus. Methodology: Our method combines two main techniques: Convolutional Neural Networks (CNN) for segmenting lesions and a technique called RootSIFT to enhance the CNN for classifying the lesions. For the segmentation task, we use a type of neural network known as CNN, which can recognize and outline the exact regions of the monkeypox lesions in images. To improve the classification performance, we introduce the RootSIFT technique. This technique enhances the features used to classify the lesions. RootSIFT is derived from SIFT (Scale- Invariant Feature Transform) key points, and we incorporate it into the CNN- based model for better identifying the distinguishing features of the lesions. Results: To test our approach, we employed a comprehensive dataset containing images of monkeypox lesions. The dataset was divided into three parts for training, validation, and testing purposes. Our experimental results demonstrate the superiority of our approach over traditional CNN methods. We achieved accurate segmentation of the lesions and improved classification accuracy as compared to conventional techniques. Conclusions: The outcomes of this research underscore the potential benefits of merging advanced image analysis methods to achieve accurate and efficient analysis of skin lesions. This approach could have valuable applications in dermatology clinics, assisting dermatologists in diagnosing skin diseases more precisely and categorizing them correctly.
Keywords:	Segmentation, classification, RootSIFT, CNN, Dermatological, Diagnosis, Image Analysis, Scale-Invariant Feature Transform.
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	BLACK WIDOW’S NEW APPROACH TO TACKLE THE TRAVELING SALESMAN PROBLEM
Author:	KHAOULA CHERRAT , RIFFI MOHAMMED ESSAID
Abstract:	This article explores the application of the Black Widow algorithm, a metaheuristic algorithm inspired by nature, to solve the classic Traveling Salesman Problem (TSP). The Black Widow algorithm is inspired by the predatory behavior of black widow spiders and uses a novel approach to guide the search for the best solution. We discuss the implementation of the algorithm and its effectiveness in handling NP-hard TSP. Through rigorous experiments and comparative analysis, we demonstrate that the algorithm can effectively navigate the solution space and produce promising results in terms of solution quality and computational efficiency. This research contributes to the continuous development of optimization algorithms and provides an opportunity to further explore bionic technology to solve combinatorial optimization problems. The effectiveness of the BWO algorithm in finding optimum solutions to the benchmark functions is tested over 51 different benchmark functions. The findings show that the suggested approach outperforms previous algorithms in several ways.
Keywords:	Traveling Salesman Problem, Black Widow Optimization, Combinatorial Optimization Problem, Black Widow Spiders.
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	ENERGY EFFICIENCY IN INDUSTRY 4.0: IMPLEMENTING REAL-TIME POWER FACTOR CORRECTION IN SMART MANUFACTURING
Author:	YOUSSEF ZERGUIT, YOUNES HAMMOUDI, MOSTAFA DERRHI
Abstract:	The Industry 4.0 paradigm has ushered in a new era of smart manufacturing, optimizing industrial processes through advanced technologies. This transformation highlights the pressing need for efficient energy utilization. Despite their crucial role in industrial power distribution, three-phase systems face persistent challenges due to load unbalances, which compromise energy efficiency and lead to suboptimal power factor levels. This unbalance results in excessive energy consumption, escalated costs, and potential equipment strain. Our research introduces an innovative methodology that harnesses Industry 4.0 technologies to reshape power factor correction and phase angle balance in three-phase systems. By strategically integrating Power Monitoring and Control Units (PMCU) within the electrical network, our approach enables real-time adjustments of capacitors and inductors. This dynamic control mechanism ensures that each load's power factor consistently approaches unity, thereby optimizing energy utilization and reducing wastage. Motivated by two key factors, our research aims to capitalize on Industry 4.0 principles for heightened adaptability and responsiveness in power systems. Moreover, the potential energy savings and operational efficiencies stemming from enhanced power factor correction have far-reaching implications for both industrial and environmental sustainability. By bridging theoretical insights with practical implementation, our work facilitates more efficient and intelligent power distribution within the Industry 4.0 landscape. In summary, our research addresses the pivotal challenge of load unbalances through an innovative methodology, contributing to the ongoing transformation of industrial processes towards enhanced efficiency, sustainability, and economic viability.
Keywords:	Automatic Balancing System, Power Factor Correction, Energy Efficiency, Three-Phase Loads, Three-Phase Balancing, Industry 4.0.
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text

Title:	IDENTIFYING INFLUENTIAL NODES IN DIRECTED WEIGHTED NETWORKS USING PYTHAGOREAN FUZZY SETS
Author:	VENKATA RAO SONGA, DR. PRAJNA BODAPATI
Abstract:	Centrality considers node importance in complex networks, addressing this issue poses a significant challenge in the realm of social network analysis. Over the recent years, various measures of centrality have been suggested to evaluate the impact of nodes inside a network. However, these measures have certain drawbacks based on the network structure , lack of ground-truth values etc. This study introduces a new centrality metric called Node Pack Fuzzy Information Centrality (NPFIC), which suggests that crucial information about a node's significance can be derived from the internal structure of its pack. NPFIC quantifies the significance of a node by assessing the information content within its pack, which is calculated by the improved Havrda and Charavat entropy. We use Pythagorean Fuzzy Sets to address the uncertainty associated with the contributions of neighboring nodes to the centrality of the center node, this is often overlooked by established traditional approaches. To illustrate the effectiveness of the proposed approach, we compare it with four established centrality measures. We conduct experiments on a real-world directed weighted complex network to validate its performance and we employed the susceptible-infected-recovered (SIR) model to assess the effectiveness of our proposed approach. The outcomes of our experiments reveal that the crucial nodes identified by NPFIC significantly influence network connectivity.
Keywords:	Directed Weighted Complex Networks, Pythagorean Fuzzy Sets (PFSs), Node Pack Fuzzy Information Centrality(NPFIC), SIR , Havrda-Charvat Entropy, Centrality Measures.
Source:	Journal of Theoretical and Applied Information Technology 15^th January 2024 -- Vol. 101. No. 1-- 2024
Full Text