		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
February 2024 | Vol. 102 No.3

Title:	UNVEILING TRUST DYNAMICS: A NOVEL EXAMINATION INTO INFLUENTIAL FACTORS OF INDONESIAN C2C SOCIAL MARKETPLACES
Author:	JUAN DANIEL WIJAYA, NILO LEGOWO
Abstract:	Facebook Marketplace is one of the biggest social-media-based marketplaces in the world, in which includes Indonesia. With many transactions occurring in Indonesia, there are different aspects that user determines when deciding whether to trust Facebook Marketplace and the sellers within. Therefore, this study aims to analyze and determine factors influencing a person's trust aspect when transacting on Facebook Marketplace. This study uses a quantitative method to collect data by distributing questionnaires to 151 respondents via Google Forms; then, the data is processed and analyzed using the Partial Least Square Structural Equation Modeling (PLS-SEM) method. The findings reveal that Credibility, Payment Options, Price, and Ease of Use is positively impacting Satisfaction. Moreover, Satisfaction, along with Privacy and Word of Mouth, positively influences Trust. However, this research shows that Security and Brand Awareness do not significantly affect Trust and may not be the key determinant of user trust. Overall, this research provides valuable insights into the factors influencing the trust of users on the Facebook Marketplace in Indonesia. The results highlight the significance of Credibility, Payment Options, Price, Ease of Use, Satisfaction, Privacy, and Word of Mouth in shaping users trust. The research results can guide Facebook Marketplace and other social marketplace platforms to improve security measures and users brand awareness, to drive users trust.
Keywords:	C2C, Social Marketplace, Social Commerce, Satisfaction, Trust, Indonesia
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	ENHANCING ANDROID MALWARE DETECTION THROUGH ENSEMBLE STACKING CLASSIFIERS AND REGULARIZATION-BASED FEATURE SELECTION
Author:	ELISA BAHARA SORITUA, DITDIT NUGERAHA UTAMA
Abstract:	With the persistent evolution of Android malware, advanced detection techniques have become paramount. This paper introduces a novel approach to Android malware detection, harnessing the power of ensemble stacking classifiers combined with regularization-based feature selection techniques, specifically Lasso, Ridge, and Elastic Net, applied to the Drebin-215 dataset. Using base classifiers including Random Forest, K-Nearest Neighbors, Support Vector Machine, Logistic Regression, and Bernoulli Naive Bayes, with Logistic Regression as the meta classifier, our methodology aims to capture the collective strengths of diverse algorithms. Initial results of individual classifiers laid the foundation, upon which the ensemble model furthered the detection rates. Implementing the regularization-based feature selection significantly enhanced the model's efficiency, leading to improved classification accuracy. Compared to traditional methods, our proposed system offers a notable enhancement in malware detection capabilities, providing a resilient solution to the prevailing challenges in Android security. This study underscores the potential of integrating machine learning-driven ensembles with advanced feature selection techniques for bolstered security measures.
Keywords:	Android Malware Detection, Machine Learning, Ensemble Stacking, Genetic Algorithm.
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	THE PERFORMANCE OF THE 3DES AND FERNET ENCRYPTION IN SECURING DATA FILES
Author:	SITI MUNIRAH MOHD, SHAFINAH KAMARUDIN, NORHANIZAH YAHYA, SAZLINAH HASAN, MUHAMMAD LUQMAN MAHAMAD ZAKARIA, SAHIMEL AZWAL SULAIMAN, DJOKO BUDIYANTO SETYOHADI
Abstract:	The number of cyber attacks launched has increased five-fold since the advent of the Coronavirus Disease Pandemic (COVID-19) in 2019. Ransomware is currently one of the highest digital risks as it aids cybercriminals to use persistent threat tools and techniques to get access to targeted networks by way of third parties. Therefore, this study aims to implement the symmetric encryption algorithms known as 3DES and Fernet methods as a means for securing files. In addition, this study evaluates the 3DES and Fernet encryption methods' performance in protecting confidential file. This study significantly contributes in comprehending Fernet and 3DES methods for securing confidential files, identifying the most efficient cryptographic symmetric algorithm for file security, and providing comparative results between Fernet and 3DES. This study applies both 3DES and Fernet in the scenario of client-server architecture in performing encryption and decryption processes. The results from the current study has shown successful implementation of these two encryption methods for both the encryption and decryption processes. In addition, this study evaluates the temporal efficiency for the encryption process. Five different text file sizes ranging from 10KB to 50KB were used for the experimental trial in evaluating the performance of both encryption methods. The outcome reveals that the Fernet encryption method performs better than 3DES.
Keywords:	Fernet Algorithm, 3DES Algorithm, Encryption, Decryption, File Security
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	SECURE AND ISOLATED COMPUTING IN VIRTUALIZATION AND CLOUD ENVIRONMENTS: A SYSTEMATIC REVIEW OF EMERGING TRENDS AND TECHNIQUES
Author:	MOHAMMAD SHKOUKANI, SAMER MURRAR, RAWAN ABU LAIL, KHALIL YAGHI
Abstract:	In an era defined by the omnipresence of virtualization and the increasing dependence on cloud computing, ensuring efficient and secure virtual machine management is of paramount importance. This paper presents a detailed review of some innovative studies, each addressing distinct challenges in the realm of cybersecurity and virtual machine performance. The first study examines the use of Lightweight Kernels and Trusted Execution Environments to optimize security isolation capabilities, demonstrating promising results for high-performance computing platforms. The second study explores the application of machine learning techniques for detecting anomalies in cloud based virtual machine resource usage, contributing to a proactive security approach. The third study presents SecFortress, an approach that enhances hypervisor security using cross-layer isolation techniques. Together, these studies underscore the significance of continual research in secure, efficient computing and offer promising avenues for future development.
Keywords:	Virtual Machine, High-Performance Computing, Cybersecurity, Machine Learning, Hypervisor Security
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	DEVELOPMENT OF PATHFINDING USING A-STAR AND D-STAR LITE ALGORITHMS IN VIDEO GAME
Author:	RISKA NURTANTYO SARBINI, IRDAM AHMAD, ROMIE OKTOVIANUS BURA, LUHUT SIMBOLON
Abstract:	Currently Artificial Intelligence (AI) is a very important component and is commonly used in video games. AI is used in games to make the player's experience more interesting and interactive. Artificial intelligence (AI) is utilized in a number of video games to enhance the player's experience and make it more interactive, particularly in role-playing games (RPG). Find a path (pathfinding) is essential to many computer games, especially role-playing games, and is required of the AI in the majority of games. Using the A* algorithm is one of the pathfinding methods used in video games to find the shortest path on the track to avoid static or dynamic obstacles, and The D* Lite algorithm is highly successful and capable of eliminating several difficulties and producing more ideal outcomes. The purpose of our research is to build and evaluate the performance of an artificial intelligence pathfinding system in searching for the fastest route using the A-star and D-star Lite algorithms. This work was carried out to collect data on how NPC movement works using path finding and study the results of this research. The path finding methods that will be studied are the A* algorithm and the D* Lite algorithm. The contribution of this research provides benefits to path finding problems which are commonly used by NPCs for technological games in the future, especially when using the A* algorithm and D* Lite algorithm in game technology. However, the pathfinding method is not only for games, but can also be implemented in other fields. The results of the experiment at five points under Pathfinding mechanism using the D-star Lite algorithm were faster in finding the closest route with a record time of 00:01.847 while using the A-star algorithm obtained 00:05.231.
Keywords:	Video Game, Artificial Intelligence, Pathfinding, A-Star Algorithm, D-Star Lite Algorithm
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	DEEP LEARNING AND STATISTICAL OPERATIONS BASED FEATURES EXTRACTION FOR LUNG CANCER DETECTION SYSTEM
Author:	SULEYMAN A. AlSHOWARAH
Abstract:	Lung cancer is considered as most common cause of cancer death than other. This kind of cancer is growing in human body without previous symptoms. So, using systems to diagnosis the patients in early stages is very essential and conducting studies in this field to find a good accuracy is also required. This research aims to examine the possibility of using of Deep Learning techniques for the lung cancer classification based on VGG-19 using images. Layer 6 and layer 7 of VGG19 were used. Also, new datasets will be created from these two layers named as statistical operations, which includes: average, minimum, maximum and combination between the two layers. Then, the datasets will be classified using different ML classifiers, which includes: KNN, Random Forest, Naïve Bayes and Decision Tree. Three scenarios will be used based on the training dataset size when classifying data. In the results, KNN scored the best accuracy (98.40), precision (0.98), recall (0.98) and F-measure (0.98). The results were nearly similar in all layers and scenarios; this means that the extracted features can provide high accuracy if applied in classification researches. It can be proved that the lung cancer can be detected with best accuracy even if the size of dataset in the training set was small. Also, the second-best accuracy after KNN algorithm is Random Forest in all layers and scenarios.
Keywords:	Lung Cancer Detection Using DL, Vgg-19, Lung Tumor, Benign Or Malignant Of Lung Cancer.
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	A METHOD FOR ENHANCING WEB SEARCH RESULTS USING CONTEXT-BASED INDEXING
Author:	MENNATOLLAH MAMDOUH MAHMOUD, DOAA SAAD ELZANFALY, AHMED EL-SAYED YAKOUB
Abstract:	While the current search engines are deemed sufficient, they often overwhelm the user with irrelevant results. This is primarily because most indexing structures fail to incorporate the context of the query. This omission significantly impacts the effectiveness of the search results. Despite extensive research is carried out to enhance search engine indexing outcomes, the problem of retrieving the most relevant results by context still exists. This study attempts to bridge this gap and contributes with a new approach to context-aware indexing, aiming to enhance the relevancy of the retrieved documents to the user s query. Unlike traditional methods that rely solely on keywords, the proposed approach leverages document context. The effectiveness of the proposed method is evaluated based on three criteria: index size, indexing construction complexity, and the relevancy of the results. Furthermore, the proposed method is compared to the Boolean retrieval model employed by the traditional information retrieval systems. The experimental results show that the proposed method outperforms the traditional information retrieval systems in terms of the indexing size and complexity, as well as the relevancy of the results.
Keywords:	Context-based Indexing, Information Retrieval (IR), Modifying Index, Search Engine, Web Document Indexing.
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	DEPTH ESTIMATION METHOD BASED ON RESIDUAL NETWORKS AND SE-NET MODEL
Author:	MOHAMMAD ARABIAT, SUHAILA ABUOWAIDA, ADAI AL-MOMANI, NAWAF ALSHDAIFAT, HUAH YONG CHAN
Abstract:	The objective of this study is to examine the problem of monocular depth estimation, which is essential for understanding a particular scene. The application of deep neural networks in generative models has resulted in significant progress in the precision and efficiency of depth estimations derived from a solitary image. However, most previous approaches have shown shortcomings in accurately calculating the depth barrier, resulting in less than optimal outcomes. Image restoration refers to the procedure of enhancing the quality of an image that has been degraded or has reduced clarity. This study introduces a novel and direct method that utilises the attention channel of the depth-to-depth network. This network has encoded elements that are useful for guiding the process of creating depth. The attention channel network consists exclusively of convolution layers and the Squeeze-And-Excitation Network (SENet). The deconvolution technique has the ability to produce images of excellent quality and can be efficiently taught with single-depth data. To enhance the acquisition of information and abilities in analysing the relationship between a colour value and its associated depth value in an image, we propose using a training method. The approach being suggested entails the utilisation of a color-to-depth network. This is enhanced by the inclusion of a loss function that is explicitly defined within the system. The combination with features derived from the hidden space. Regarding our attention channel network. One notable advantage of the suggested methodology is valuable due to its capacity to enhance local capabilities. To a large extent. Precise and thorough information can be Attained even in locations with intricate surroundings. The outcomes The data used in the study was derived from the NYU Depth v2 benchmark. The dataset showcases the effectiveness and durability of the proposed solution. Methodology in contrast to current cutting-edge approaches.
Keywords:	Deep Learning, Depth Estimation, Resnet-101, Features Map, NYU Depth v2.
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	ENHANCING COLLABORATIVE FILTERING: ADDRESSING SPARSITY AND GRAY SHEEP WITH OPPOSITE USER INFERENCE
Author:	ABDELLAH EL FAZZIKI, YASSER EL MADANI EL ALAMI, JALIL ELHASSOUNI, MOHAMMED BENBRAHIM
Abstract:	Collaborative filtering (CF) is a popular recommendation approach which seeks to find similar users to predict what an active user might like. However, CF suffers from two main challenges: sparsity and gray sheep. In both cases, recommending useful items is a difficult task. In this paper, we propose a new approach to address these challenges. It consists of combining Singular Value Decomposition and Association rule methods with enriched rating matrix. In addition to actual users, this matrix incorporates virtual users inferred from opposing ratings provided by real users. Our approach attempts to increase the density of similar users and makes it easier to make useful recommendations. We conducted a comparative study showing that our method outperformed traditional CF approaches in terms of accuracy.
Keywords:	Recommendation System; Collaborative Filtering; Opposite Preferences; Model-Based CF; SVD; Association Rules; Sparsity Problem.
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	DIABETIC MELLITUS PREDICTION WITH BRFSS DATA SETS
Author:	MARWA HUSSEIN MOHAMED, MOHAMED HELMY KHAFAGY, NESMA MOHAMED MAHMOUD KAMEL, AND WAEL SAID
Abstract:	One of the chronic diseases that affect many people worldwide is diabetic mellitus. If the disease is predicted at an early stage, the risk and severity can both be significantly decreased. In this research, we need to predict the type 2 diabetic patients at an early stage to reduce the cost of treatment for countries because this is a long time disease we use many machine learning algorithms to find the accuracy for these diseases applied to BRFSS datasets for two years 2014 and 2015 with a different selection of features to predict the disease as decision tree, logistic regression, ADA Boost Classifier, extreme gradient boosting, Linear Discriminant Analysis, Light Gradient Boosting Machine, and catboost classifiers. While applying our experiments with the 2014 BRFSS data sets Neural network has the highest accuracy with 82%and with the 2015 BRFSS datasets the best accuracy model was 86% for CatBoost Classifier and Extreme Gradient Boosting where the lowest model was Linear Discriminant Analysis. Also, in our research we compare our results with others using the same datasets with different features selection and get high accuracy.
Keywords:	Chronic Diseases; Diabetic Mellitus; Machine Learning; Artificial Intelligence; Classification Models.
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	PRUNING-BASED HYBRID NEURAL NETWORK FOR AUTOMATIC MODULATION CLASSIFICATION IN COGNITIVE RADIO NETWORKS
Author:	NADIA KASSRI, ABDESLAM ENNOUAARY, SLIMANE BAH3, IMAD HAJJAJI, HAMZA DAHMOUNI
Abstract:	Today, there is a greater need than ever to focus on realistic solutions to efficiently manage the available frequency spectrum, due to the scarcity of this limited resource. Thanks to its opportunistic communication paradigm allowing fair access for all users, Cognitive Radio (CR) has been commonly recognized as an enabling technology to mitigate the spectrum scarcity problem and ensure optimal use of the available spectrum. Automatic Modulation Classification (AMC) is an indispensable function for carrying out various CR tasks, including link adaptation and dynamic spectrum access. Recently, Deep Learning (DL) networks have shown promising results in dealing with AMC compared to traditional methods. However, most DL-based approaches proposed so far have high computational and memory requirements, which make them impractical for resource-constrained IoT applications. To address this challenge, we introduce in this work a novel lightweight hybrid Neural Network for AMC, consisting of a combination of Gated Recurrent Unit (GRU) and Convolutional Neural Network (CNN) structures, in which the fully connected layers are omitted. Our proposed design outperforms the baseline models in terms of recognition accuracy across various Signal-to-Noise Ratios (SNRs), while maintaining a lightweight structure and low computational complexity. To further enhance model compression and resource utilization, we have introduced an Iterative Magnitude-Based Pruning approach combined with Quantization Aware Training (QAT). Simulation results using the RadioML 2018.01A dataset validate the efficacy of our proposed model. With a sparsity of 0.7, the pruned variant, comprising 80 GRU cells, attains an average accuracy of 60.88% across all SNRs. Remarkably, it achieves a highest accuracy of 95.95% at 20 dB, while utilizing only 8 210 parameters.
Keywords:	Automatic Modulation Classification, Convolutional Neural Network, Gated Recurrent Unit, Cognitive Radio Networks, Spectrum Sensing, Resources-Constrained Objects, Pruning, Quantization Aware Training.
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	ENHANCED INSIDER THREAT DETECTION THROUGH MACHINE LEARNING APPROACH WITH IMBALANCED DATA RESOLUTION
Author:	PENNADA SIVA SATYA PRASAD, SASMITA KUMARI NAYAK, Dr. M. VAMSI KRISHNA
Abstract:	An insider threat is the risk that person inside an organization may pose to the company's security, data, or resources. Insider threat detection is a crucial component of a comprehensive cybersecurity strategy. By identifying and mitigating risks from within the organization, businesses can better protect their assets, maintain trust, and ensure compliance with legal and regulatory requirements. This paper addresses the detection of insider threats using machine learning algorithms. A famous CERT dataset was used for the experiments. The collected dataset is largely imbalanced. The ML algorithms cannot perform well with imbalanced datasets. So, data imbalance can be resolved by using three over sampling techniques namely random oversample, smote, adasyn and three under sampling techniques namely random under sample, Cluster centroids and Edited Nearest Neighbors. Later, five ML algorithms namely Logistic Regression, Adaboost, Decision Tree, Random Forest and Naïve Bayes applied to the datasets generated through over sampling and under sampling techniques. To further increase the performance of the model, an ensemble learning is proposed along with principal component analysis. The experimental results demonstrated that the proposed model surpassed the performance of existing models for insider threat detection.
Keywords:	Insider Threat Detection, Data Imbalance, Over Sampling, Under Sampling, Machine Learning, Ensemble Learning.
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	PROVISIONING AN UNCERTAINTY MODEL FOR ELEPHANT MOVEMENT PREDICTION USING LEARNING APPROACHES
Author:	R. VASANTH, A. PANDIAN
Abstract:	This work concentrates on resolving Static/Dynamic Body movement estimation and rigid body orientation of animals. Algorithm has to be modeled with a complementary structural model that exploits measurements of the magnetometer, accelerometer, and gyroscope. Usually, attitude information is essential to evaluate animals' movement estimation to compute active sensor nodes to gather information. The risk factor is measured with Weighted Uncertainty Priority (WUP), an IVF framework. WUP considers various relative weights of complex risk factors by examining their degree of ambiguity/uncertainty. Uncertainty measure is an evidence theory used to generate exponential weight of every risk factor. This ambiguity measure shows a subjective assessment of internal coordination between sensors and animal movements. The anticipated algorithm reduces network overhead and effectually classifies samples with the pre-trained Yolov5 Network model to estimate the arrival of animals and to determine the correlation between normal and uncertain data. Feature-aware pattern modeling is done to schedule elephant movement versus risk factors. The theoretical analysis has to validate that the anticipated model outperforms other prevailing models. Simulation has been carried out in a MATLAB 2020a environment where detection accuracy is estimated as 99.5%.
Keywords:	A Priority-Based Weighted Scheduling, Weighted Uncertainty Priority, Non-Linear Filter, Uncertainty, Kalman Filter, Static/Dynamic Body Movement Estimation, Rigid Body Orientation
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	MINING DEVIATION WITH MACHINE LEARNING TECHNIQUES IN EVENT LOGS WITH AN ENCODING ALGORITHM
Author:	DR. V.V. JAYA RAMA KRISHNAIAH, DR BANDLA SRINIVASA RAO, DUGGINENI VEERAIAH, MR. S. SUBBURAJ, DR. MOHAMMED SALEH AL ANSARI, DR. CHAMANDEEP KAUR
Abstract:	A study field called "commercial operation divergence analysis" tries the identify how a commercial system varies beyond the results which were anticipated. Approaches in this field identify the qualities of a collection of procedure executes may are related to shifts in process efficiency, exposing the characteristics of procedure behaviours which generate undesirable procedure for the process as well as understandings of what process behaviour contributes the greatest efficiency. Success in this scenario can relate to any domain-dependent efficiency measurement as well as the expense, time, and resources factors. The finding of trends from the logs of events using various trend mining approaches is the basis of the present-day company deviation mining methodologies. Such extractor methods are now provide to a small degree of flexibility because they're unable to represent the intricate linkages which could be present in highly variable systems. Within this paper, then offer an innovative decoding strategy for vector-based representations of procedure scenarios, followed by then utilise Machine Learning for a unique approach in the context of Deviance Mining to pinpoint the aspects of a procedure which most significantly affect its efficiency. The outcomes demonstrated how machine learning delivered pertinent as well as emotive conclusions on the event's logs when combined with the suggested Declare-based coding, constituting an effective tool for the analysis of processes.
Keywords:	Machine Learning, Mining Deviation, Encoding, Logs
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	ANALYZING LARGE SCALE BUSINESS DATA TO PREDICT COMPANY'S GROWTH USING AN INTEGRATED HYBRID APPROACH OF DATA REDUCTION UNIT AND CONVOLUTIONAL NEURAL NETWORKS
Author:	KATHARI SANTOSH, DR. G. SUNDAR, DR.P. PATCHAIAMMAL, SWETHA V P, BADUGU SURESH, DR. SANTHOSH BODDUPALLI
Abstract:	In current quickly changing business world, large-scale business data analysis has become critical for predicting the development of a company as well as taking effective decisions. The paper proposes an integrated hybrid strategy that leverages the benefits of data reduction units and convolutional neural networks (CNNs) to reliably estimate the growth potential of a company. The strategy's initial component is data reduction units, which minimize the complexity of data while keeping informative quality. The strategy's second component makes use of the ability of convolutional neural networks, a deep-learning framework suitable for processing both organized and unstructured input. CNNs excel at collecting geographical and time-related trends in data, making them excellent for finding complicated links in large-scale corporate information. The system is capable of learning key characteristics and hierarchies by utilizing many layers of convolutional filters, allowing it to generate accurate growth predictions. The work proposes a unique architecture for integrating these components, which mixes the outputs of the data reduction unit with the input data and passes them into the CNN. The combined strategy lets the CNN to concentrate on the most important information, improving prediction accuracy while lowering computational overhead. The study conducted tests with real-world company datasets to assess the efficiency of the strategy. The findings show that the integrated hybrid strategy exceeds standard methodologies in terms of accuracy in predicting company development. Furthermore, the study demonstrates that the technique is adaptive and scalable, with the ability to handle large-scale datasets.
Keywords:	Big Data, Convolutional Neural Network (CNN), Data Reduction, Deep Learning
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	BUILDING A MODEL FOR AUTOMOTIVE SOFTWARE DEFECTS PREDICTION USING MACHINE LEARNING AND FEATURES SELECTION TECHNIQUES
Author:	RAMZ TSOULI FATHI, MAROI TSOULI FATHI, MOHAMMED AMMARI, LAÏLA BEN ALLAL
Abstract:	Over the past few decades, the integration of software into vehicles has experienced exponential growth. This expansion affects nearly all aspects of recent automotive engineering, encompassing various components. Simultaneously, testing becomes increasingly crucial. Significant efforts are required for software verification and validation to meet security, quality, and reliability standards. These essential tests become more complex and costly with the proliferation of functionalities, risking delays in market deployment. The integration of new technologies, such as predictive analysis based on machine learning algorithms, could anticipate the expected number of anomalies, improving resource management, reducing deployment time, and mitigating the risk of potentially disastrous software errors. This paper aims to assess the applicability of software defect prediction methods in the automotive domain. To achieve this, we will construct a dataset from real industrial software projects, anonymized and confidential. This effort will culminate in the creation of a new and unique database that will serve as the foundation for our study. Through a series of experiments, we will evaluate the relevance of various machine learning algorithms, aiming to surpass classical approaches in constructing our predictive analysis model. This paper introduces a novel solution for predicting software faults in the automotive domain. The innovation lies in the realistic approach, relying on the application of complex domain knowledge to a unique database derived from real industrial automotive software projects. Systematic efforts were exerted not only to optimize the usability of this dataset but also to achieve superior performance.
Keywords:	Machine Learning, Features Selection, Predictive Analysis, Automotive Software, Software Metrics, Automotive Software Projects, Software Defect Prediction
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	DYNAMIC ACCESS CONTROL AT THE NETWORK EDGE USING AN ADAPTIVE RISK-BASED ACCESS CONTROL SYSTEM (ad-RACs)
Author:	MUHAMMAD BELLO ALIYU, DR. MUHAMMAD GARBA, DR. DANLAMI GABI, DR. HASSAN U. SURU, DR. MUSA S. ARGUNGU
Abstract:	The widespread adoption of edge computing models owes to their cost-effectiveness and performance advantages for both users and service providers. However, the expanding user base and application scope raise security concerns, including potential malicious attacks due to unrestricted system resource access. Hence, this study focuses on implementing an Adaptive Risk-based Access Control System (ad-RACs) at the network edge. The ad-RACs utilizes four key inputs—user context, resource sensitivity, action severity, and risk history—to enable the CatBoost risk estimation module to evaluate security risks associated with access requests. Upon meeting the acceptable risk threshold, the Chinese wall access control policy determines access decisions. This model adapts to user behavior and patterns, updating risk history to dynamically adjust access requests. Evaluation results showed that the ad-RACs exhibited satisfactory recall and F1 score values of 100% and 98%, respectively, and a precision value of 95%, outperforming the existing system's recall of 98%, F1 score of 96%, and precision of 97%. Conclusively, the ad-RACs excelled in recall and F1 score values compared to the existing system, indicating its potential to enhance access control. Its adoption is recommended for governmental and private organizations seeking to bolster user access to sensitive resources.
Keywords:	Access Control, Adaptive Access, Adaptive Security, Network Edge, Risk-based Access
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	DECODING ADVERSARIAL MACHINE LEARNING: A BIBLIOMETRIC PERSPECTIVE
Author:	JEENA JOSEPH, JOBIN JOSE, DIVYALAKSHMI S, RAJIMOL A, GREETY TOMS, ANAT SUMAN JOSE, GILU G ETTANIYIL
Abstract:	The rapid improvement of machine learning techniques has led to an extraordinary rise in the prominence of adversarial attacks and their accompanying defenses. This study performs a comprehensive bibliometric analysis of adversarial machine learning, providing insight into the field's evolution from its foundation to the present. We identify the primary themes, foundational works, and influential figures in this field using state-of-the-art bibliometric technologies and databases. Our findings demonstrate the impressive growth of adversarial machine learning research and emphasize its transdisciplinary nature. We also highlight the collaborative networks and important hubs that have fueled advancements in this discipline. This report offers a thorough perspective on adversarial machine learning, its major turning points, and valuable insights for researchers, educators, and practitioners.
Keywords:	Adversarial Machine Learning, Bibliometric Analysis, VOSviewer, Biblioshiny.
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	BLOCKCHAIN IN LAND REGISTRY FOR TRANSFORMING LAND ADMINISTRATION
Author:	VISHNU SHUKLA, Dr. ABHIJEET R RAIPURKAR, Dr. MANOJ B CHANDAK, VEDANT BARAI
Abstract:	Land registry systems play a critical role in recording and verifying property ownership, ensuring transparency, and facilitating secure transactions. However, traditional land registry systems often face challenges related to inefficiencies, fraudulent activities, and lack of trust. The emergence of blockchain technology has the potential to address these issues and revolutionize land registries. This paper present blockchain-based land registry system that leverages the unique features of blockchain technology to provide a secure, transparent, and tamper resistant platform for recording and managing land ownership information. By utilizing distributed ledger technology, the proposed system aims to eliminate the need for intermediaries, reduce the risk of fraud, enhance data integrity, and streamline the land registration process. The blockchain-based land registry system would involve the creation of a decentralized network where each participating node stores a copy of the registry, ensuring redundancy and immutability. The use of cryptographic techniques would provide secure and verifiable transactions, allowing for a trustworthy and auditable history of land ownership transfers. A blockchain-based land registry system has the potential to transform traditional land registry practices by providing a secure, transparent, and efficient platform for recording and managing land ownership information. By leveraging the decentralized and immutable nature of blockchain technology, this system can significantly enhance trust, streamline processes, and reduce fraud in land transactions, leading to more reliable and effective property markets.
Keywords:	Blockchain, Land Registration, Smart Contracts, Solidity, Ownership-Transfer, Migration, Map, DAPP.
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	NUMBER PLATE AND LOGO IDENTIFICATION USING MACHINE LEARNING APPROCHES
Author:	SYED ARFATH AHMED, PROF (DR) SARVOTTAM DIXIT
Abstract:	The ability to recognize license plates is affected by a number of issues, such as the presence of extraneous information, variations in the size and type of the text, blur, skew, and other environmental factors. It is not always easy to recognize the numbers on license plates because there are so many different types of licence plates and so many different situations in which they can be displayed. As a result of the relatively poor segmentation results produced by region-based approaches, better segmentation algorithms are required for the process of separating out individual license plates. The license plate can be placed in any part of the image that you like. Because the license plate can be identified based on its attributes, the algorithm processes only the pixels that share analogous qualities. The amount of time required to process the image will increase if each and every pixel in it is processed. In this study, two additional methods that are proven to be effective in segmenting the license plate of a vehicle are investigated. The first technique divides the license plate by utilizing edge information. This segmentation technique demonstrates a high level of effectiveness, providing an average segmentation outcome of 85%. The second approach utilizes information about number plate shape and color. Visualization techniques are applied to segment number plates using the information of the hybrid feature. The approach achieves an average segmentation rate of 93%. Although segmentation using the edge-based and hybrid approaches is executed at a similar speed to segmentation using the region-based approach, the latter provides better segmentation outcomes. The fundamental objective of this paper is to devise a method that is capable of producing an automatic licence plate recognition system that is durable, accurate, and reliable.
Keywords:	Automatic Number Plate Recognition (ANPR), Convolution Neural Network (CNN), Character Segmentation Image Segmentation, Optical Character Recognition, Edge & Pixel Detection.
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	DEVELOPMENT AND EVALUATION OF PNEUMFC NET: A NOVEL AUTOMATED LIGHTWEIGHT FULLY CONVOLUTIONAL NEURAL NETWORK MODEL FOR PNEUMONIA DETECTION
Author:	SHUBHRA PRAKASH, B RAMAMURTHY
Abstract:	The aim of this study is to address the challenges of pneumonia diagnosis under constraint resources and the need for quick decision making. We present the PneumFC Net, a novel architectural solution where our approach focuses on minimizing the number of trainable parameters by incorporating transition blocks that efficiently manage channel dimensions and reduce number of channels. In contrast to using fully connected layers, which disregard the spatial structure of feature maps and substantially increase parameter counts, we exclusively employ only convolutional layer approach. In the study, X-ray image dataset is used to train and evaluate the proposed Convolutional Neural Network model. By carefully designing the architecture, the model achieves a balance between parameters and accuracy while maintaining comparable performance to pre-trained models. The results demonstrate the model's effectiveness in detecting pneumonia images reliably. In addition, the study examines the decision-making process of the model using Grad-CAM, which helps to identify important aspects of radiographic images that contribute to the positive pneumonia prediction. Furthermore, the study shows that the proposed model, Pneum FC Net not only has the highest accuracy of 98%, but the total trainable model parameters is only 0.02% of the next best model VGG-16, thus establishing the potential of this new robust Deep Learning model. This research primarily addresses concerns related to mitigating significant computational requirements, with a specific focus on implementing lightweight networks. The contribution of this work involves the development of resource-efficient and scalable solution for pneumonia detection.
Keywords:	PneumFC Net, Fully Convolutional Neural Network, Pneumonia Detection, Computer Aided Diagnosis
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	OPTIMIZING SENTIMENT ANALYSIS IN ONLINE SHOPPING: UNVEILING THE PRECISION OF SIMPSON RULE-OPTIMIZED SUPPORT VECTOR MACHINE
Author:	R. ANITHA, D. VIMAL KUMAR
Abstract:	In the domain of online shopping, customer reviews serve as invaluable repositories of insights, encapsulating the sentiments and experiences associated with a wide array of products. However, the extraction of nuanced sentiments from this extensive pool of reviews faces a formidable challenge due to the inherent complexities of language and context. Sentiment analysis, a pivotal tool for distilling sentiments from textual data, grapples with accurately deciphering nuanced expressions, sentiment subtleties, and contextual intricacies. To address these challenges, this study introduces the Simpson Rule-Optimized Support Vector Machine (SR-SVM) for sentiment analysis in online shopping. Built on Support Vector Machines (SVM) principles, SR-SVM leverages Simpson's rule to optimize sentiment pattern identification within the expansive landscape of online shopping reviews. Through the application of mathematical optimization techniques, SR-SVM refines sentiment analysis outcomes, promising a more nuanced understanding of customer sentiments. Preliminary results indicate a noteworthy enhancement in sentiment classification accuracy, underscoring the transformative potential of SR-SVM in optimizing sentiment analysis for online shopping platforms. This study opens a promising avenue for refining and advancing sentiment analysis methodologies in the dynamic and complex context of online retail.
Keywords:	Sentiment, Classification, Amazon, Simpson Rule, SVM, Optimization
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	LEARNING OF ENERGY EFFICIENT AND NETWORK TRAFFIC DELAY IN WIRELESS NETWORKS USING CHANNEL AWARE ROUTING
Author:	K. BALASUBRAMANIAN, R. MANIVANNAN, S. MANIKANDAN, A. HAJA ALAUDEEN, MISHMALA SUSHITH, KIRUTHIGA BALASUBRAMANIYAN
Abstract:	This paper studies different analysis metrics of wireless networks using channel aware routing protocol. During communication the following characteristics are important performance metrics like network traffic delays and energy efficient. Energy Efficient is a important solution with limited buffer capacity and life time of battery in wireless environment. The status of each node can be identified by using co-channel forwarding mechanism. The traffic delays can be measured by using Q-Learning random early method and routing can be optimized by channel aware routing. The transmission changes in one node to another can be monitored by router database and energy level can be verified by additional changes in wireless network nodes. In this paper, we formulate to achieve analysis the energy performance and reduced traffic delay using co-channels stochastic optimization procedure. The co-channel method used to check various analytical bounds and check all the nodes are uniformly shares the energy level. The mathematical analysis report provides uniform distribution energy values and traffic delays. The simulation shows efficient energy utilization and network traffic delays in wireless networks.
Keywords:	Wireless Networks, Energy Efficient, Network Traffic Delays, Channel Aware Routing, Routing Protocol.
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	CHOOSING THE RIGHT CHAOTIC MAP FOR IMAGE ENCRYPTION: A DETAILED EXAMINATION
Author:	SAAD A. ABDULAMEER
Abstract:	This article investigates how an appropriate chaotic map (Logistic, Tent, Henon, Sine ...) should be selected taking into consideration its advantages and disadvantages in regard to a picture encipherment. Does the selection of an appropriate map depend on the image properties? The proposed system shows relevant properties of the image influence in the evaluation process of the selected chaotic map. The first chapter discusses the main principles of chaos theory, its applicability to image encryption including various sorts of chaotic maps and their math. Also this research explores the factors that determine security and efficiency of such a map. Hence the approach presents practical standpoint to the extent that certain chaos maps will become relevant toward implementing image encryption system. This helps them select the best chaotic map for image encryption to ensure secure digital data.
Keywords:	Image Encryption, Chaotic Map, Chaos-based, Image Security, Cipher
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	DYNAMIC THRESHOLD GENERALIZED ENTROPIES-BASED DDOS DETECTION AGAINST SOFTWARE-DEFINED NETWORK CONTROLLER
Author:	HUSSEIN A. AL-OFEISHAT
Abstract:	This study presents a new strategy that combines dynamic thresholding with generalized entropy approaches to identify Distributed Denial-of-Service (DDoS) attacks in Software-Defined Networking (SDN) environments. This research intends to address the limitations of traditional detection approaches, such as high false-positive rates and restricted adaptability. We conducted thorough testing in various simulated SDN scenarios to assess the efficacy of our approach compared to existing methodologies. The findings exhibited a substantial enhancement in both the precision of detection and the decrease in false positive occurrences, signifying a remarkable progression compared to existing techniques. This research not only fills an important void in the realm of SDN security but also lays the foundation for more adaptable and efficient ways for detecting DDoS attacks. The results have practical implications for improving network security by providing a strong answer to the changing danger of DDoS attacks in intricate network environments. In summary, this study offers a fresh viewpoint to the field of SDN security research, proposing a possible change in DDoS detection methods towards more flexible and entropy-driven methodologies.
Keywords:	Network; SDN Controller; Telecommunications Infrastructures; Renyi; Networking; SDN.
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	OPTIMAL APPROACH FOR MINIMIZING DELAYS IN IOT-BASED QUANTUM WIRELESS SENSOR NETWORKS USING NM-LEACH ROUTING PROTOCOL
Author:	J. RAMKUMAR, A. SENTHILKUMAR, M. LINGARAJ, R. KARTHIKEYAN, L. SANTHI
Abstract:	IoT-based Quantum Wireless Sensor Networks (IoT Q-WSN) merge Quantum principles into the realm of Wireless Sensor Networks, introducing intricate routing challenges that demand innovative solutions for efficient data transmission. This study introduces NM-LEACH, an inventive routing protocol inspired by the leadership principles of Shri Narendra Modi. A groundbreaking feature of NM-LEACH is its distinction as the inaugural optimization protocol inspired by a human personality. NM-LEACH operates through adaptive strategies, clean coding practices, and continuous feedback loops, embodying a comprehensive and disciplined approach to network development. Through simulation in NS3, the protocol undergoes meticulous evaluation against existing counterparts. Results demonstrate NM-LEACH s superior performance in minimizing delays and optimizing data transmission within IoT Q-WSNs. This research advances Quantum IoT and underscores the potential of drawing inspiration from human leadership qualities to innovate and enhance wireless sensor network functionality by minimizing delay and energy consumption.
Keywords:	IoT, WSN, Quantum, LEACH, Narendra Modi, Energy Consumption
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	PROVING THE CAR SECURITY SYSTEM MODEL USING CTL AND LTL
Author:	RAFAT ALSHORMAN
Abstract:	One of the most important legal issues that troubles the courts is car theft. Therefore, car companies race to build security systems to reduce this problem. In this research, we propose a system based on remote control and a keypad to decrease the risk of car theft. To prove the system, firstly, we described it using a finite-state machine. Secondly, a set of correctness conditions are introduced. These correctness conditions are encoded in temporal logic CTL and LTL. Lastly, the NuSMV model checker is used to verify the correctness conditions. The importance of the proposed system verification is to ensure that the system is correct under these conditions. Additionally, this research also found that CTL and LTL can be used to specify and verify systems of this kind. The main contribution of this research is to demonstrate that CTL and LTL model checking can be used to specify and verify the proposed system. Moreover, this can be a stepping-stone to prove similar systems that their behaviors generate complex finite-state machines and it is difficult to trace all possible states of the system.
Keywords:	CTL, model checking, NuSMV, LTL, Car Security.
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	OPTIMIZING WATER DESALINATION: A NOVEL FUSION OF EXTREME LEARNING MACHINE AND GAME THEORY FOR ENHANCED PH PREDICTION - UNVEILING REVOLUTIONARY INSIGHTS
Author:	DR. MOHAMMED SALEH AL ANSARI
Abstract:	The essential step in guaranteeing an adequate supply of water is water desalination. Nevertheless, since many water quality measures interact in a complex way, it is difficult to estimate pH levels accurately. To further comprehend and regulate pH levels in water from desalination, this study proposes a synergistic structure that blends the predictive ability of Extreme Learning Machines (ELM) with the strategic insights offered by game theory. The precision needed for effective desalination is frequently lacking in current pH prediction techniques. Traditional models could find it difficult to represent the complex interactions between changes in pH and input factors. Furthermore, the lack of a strategic decision-making component in dynamic operational contexts exposes processes to less-than-ideal results. The merging of game theory and ELM is innovative. Because of its quick training and ability to generalize, ELM is a good pH predictor. Concurrently, Game Theory is utilized to simulate strategic exchanges between stakeholders, taking into account how pH forecasts affect decision-making procedures. High-quality data on water quality is used to train an ELM model as part of the suggested methodology. Next, using the concepts of game theory, stakeholders' strategic behaviors that are impacted by the anticipated pH levels are modelled. This research enhances pH prediction accuracy using Extreme Learning Machine and Game Theory, optimizes desalination resource allocation for sustainability, addresses real-world challenges, and provides a versatile framework for widespread application in diverse desalination scenarios. The suggested method's efficacy is determined by a thorough performance review. The integrated strategy is thought to be superior to traditional approaches in terms of metrics like accuracy of predictions, satisfaction with stakeholders, and operational effectiveness. These metrics present a potential path forward for the advancement of water desalination procedures.
Keywords:	Extreme Learning Machine, Water Desalination, Game Theory, pH prediction
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	DEEP NEURAL NETWORKS FOR EFFICIENT CLASSIFICATION OF SINGLE AND MIXED DEFECT PATTERNS IN SILICON WAFER MANUFACTURING
Author:	HUSSEIN YOUNIS, AMJAD RATTROUT, MOHAMMED YOUNIS
Abstract:	Semiconductor wafer manufacturing is a complex and costly process with inherent defect risks that can significantly impact the industry. Utilizing Deep Learning (DL) for wafer defect classification offers benefits such as improved performance, reduced human error, and time savings. This paper presents an advanced DL-based approach for wafer defect classification, based on a modified GoogLeNet model and data augmentation technique. The approach achieves state-of-the-art results on the WM-300K+ wafer map dataset, demonstrating robustness to image noise and variation. Our research introduces a pioneering approach that outperforms previous methodologies, integrating auto-cast and CUDA to enhance efficiency, addressing dataset imbalance through innovative data augmentation, and creating a new "WM-300K+ wafer map [Single & Mixed]" dataset. The methodology yields exceptional results, with an average classification accuracy of 99.9% for both single and mixed defect types, surpassing previous studies. Hyperparameter tuning with Optuna and a patient stop mechanism further fortifies the robustness and reliability of the approach.
Keywords:	Semiconductor Wafer Manufacturing, Deep Learning, Convolutional Neural Network, Wafer Defect Patterns, GoogLeNet
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	STRENGTHENING QA SYSTEM ROBUSTNESS: BERT AND CHARACTER EMBEDDING SYNERGY
Author:	RACHID KARRA, ABDELALI LASFAR
Abstract:	The inherent nonlinearity of neural networks renders them vulnerable to adversarial attacks. As artificial intelligence-based question-answering (QA) systems continue to proliferate across sectors like education and health, ensuring their security and predictable behavior becomes imperative for widespread adoption among the general public. Robustness measures the resilience of Natural Language Processing (NLP) models against adversarial attacks. A robust QA system exhibits resilient behavior when encountering maliciously crafted questions. Previous experiences have indicated that utilizing character embeddings enhances resistance against contradictory misspelled questions. To validate this, we fine-tuned BERT with character embeddings on SQuAD dataset for question-answering tasks and assessed both models using {question, context, answer} tuples. The results unequivocally demonstrate that BERT with character embeddings yields superior performance. Finally, we propose a comprehensive framework aimed at safeguarding question-answer type dialogue systems.
Keywords:	Adversarial Attacks, BERT, Character Embedding, QA Systems, SQuAd
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	DETERMINANTS OF QRIS USAGE AS A DIGITAL PAYMENT TOOL FOR MSMEs
Author:	KHALISA MARSAHANDA RAFIANI, ROCHANIA AYU YUNANDA, TOTO RUSMANTO
Abstract:	The development of technology and information in the globalization era contributes significantly to the growth of the digital economy in Indonesia. It has significant impacts on all types of industries including MSMEs (Micro, Small, and Medium Enterprises). The development of MSMEs is currently increasing in various regions throughout Indonesia. The utilization of technology by MSMEs is on the rise. Various digital technologies are being employed by MSMEs, such as electronic money and non-cash payments are gaining momentum with the introduction of QR-code payment systems. Using the Technology Acceptance Model, this study aims to examine the determinants of QRIS as a digital payment tool used by MSMEs. This study analyzes 162 from MSMEs players. Out of the six hypotheses tested, four were found to be significant or accepted. Perceived Usefulness, Perceived Ease of Use, Revenue, and Perceived Risk have significant influences on the Interest in using QRIS.
Keywords:	Technology Acceptance Model, Financial Literacy, QR Payment
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	A NETWORK ANALYTICS APPROACH TO EXAMINE THE RELATIONSHIP BETWEEN LEARNING REFLECTION AND SELF-REGULATED LEARNING SKILLS
Author:	SAHAR ALQAHTANI
Abstract:	Previous studies have demonstrated the effectiveness of Epistemic Network Analysis (ENA) to explain students' epistemic interaction with specific learning activities or tasks. However, the potential of ENA has not been widely explored in investigating the relationship between students' self-regulated learning skills and their reflective behaviors in a new learning experience. This paper demonstrates how ENA and cluster analysis can reveal and analyze differences in the reflective behaviors of groups of students with varying self-regulated learning constructs. The results of this study show that the most prominent reflections among students with a high level of self-regulation use positive feeling about their good experience and try to overcome their obstructing feelings that hinder their learning process. The following are the learning constructs: intrinsic/extrinsic goal orientation, task value, expectancy beliefs, self-efficacy, test anxiety, metacognitive awareness and metacognitive writing strategies. By contrast, students with low self-regulation in these learning constructs more frequently reflected by recollecting their negative feelings and examining the knowledge obtained from the course. The analytical approaches proposed in this study reveal that the reflective behaviors among students with both high and low motivation to learn through “intrinsic goal orientation”, “expectancy beliefs” and “self-efficacy contain no negative feelings towards their learning experience.
Keywords:	Epistemic Network Analysis, Model Graph-Based Analysis, Self-Regulated Learning, Reflection, Reflective Writing, Reflective Practice.
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	CSF AND INFLUENCE OF ITS ON 4D EDUCATION: SYSTEMATIC LITERATURE REVIEW
Author:	JAKA SUWITA, HARJANTO PRABOWO, HARCO LESLIE HENDRIC SPITS WARNARS, MEYLIANA
Abstract:	Computer literacy has evolved to become more integrated with AI approaches as technology advances, enabling the development of an adapted education system. Intelligent Tutoring System (ITS) is the name of this system. ITS is a system of tutoring that uses technology to provide assignments without the need for an educator to be there. A systematic literature review( SLR) was conducted in this study to identify the factors of an intelligent tutoring system, the determining factors for the success of ITS perpetration in the 4d-education framework, and the results are presented. The system used in this literature review is the prism system. The Prisma system uses the SLR protocol to search, opt for literature, recapitulate substantial results, and propagate results. The SLR protocol consists of determining a research question, searching for scientific reference sources with the keyword intelligent tutoring system, sorting out the appropriate articles, and making conclusions. This article provides a detailed overview of how the critical success factors and influence of ITS are related to 4D education. Our contribution is in the form of complete information regarding the ITS components that influence 4d-education related to the factors that determine the successful use of IT in higher education to improve learning and practical use of this technology.
Keywords:	Components, SLR, Its, Intelligent Tutoring System.
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	CREDIT CARD APPROVAL PREDICTION: A SYSTEMATIC LITERATURE REVIEW
Author:	INDRAJANI SUTEDJA, JACKY LIM, ERICK SETIAWAN, FRANS REXY ADIPUTRA
Abstract:	Credit cards have become an everyday payment choice, leading to a surge in applications and posing challenges for swift approval processes. This study aims to find the best machine learning algorithms for credit card approval by searching 208 papers from various sources then through inclusion and exclusion, are narrowed into 23 most relevant papers. The highlighted issues include time-consuming manual approval, high use of human resources, and possible human errors in the process. The manual method's inefficiencies can lead to wrong approvals, which could cause financial losses and harm the banks' reputation, especially if it happens in large scales. In this situation, machine learning can offer a solution for faster approval procedures and fewer errors. To address these concerns, this research evaluates different machine learning algorithms, including decision trees, random forests, logistic regression, support vector machines, and artificial neural networks. The evaluation considers algorithm ranking, statistical measures, and the nature of the algorithms to understand their effectiveness and potential for overfitting. The author's findings emphasize that adopting machine learning algorithms like random forests, logistic regression, and support vector machine can significantly enhance credit card approval processes. These algorithms exhibit higher accuracy and a reduced risk of overfitting, contributing to overall bank performance improvement. Furthermore, recommendations are provided for a more effective model development process, including choosing suitable methodologies, exploring data, reducing complexity, tuning parameters, and validating results. By following these suggestions, banks can enhance model performance while minimizing overfitting concerns.This systematic review not only underscores the importance of embracing machine learning for credit card approval but also offers practical insights into selecting algorithms and refining model development strategies. By embracing advanced algorithms and improved model building techniques, banks can navigate challenges posed by increasing credit card applications and establish more efficient, accurate, and dependable credit approval systems.
Keywords:	Credit Card, Credit Card Approval, Manual, Machine Learning
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	REVOLUTIONIZING HISTORICAL DOCUMENT ANALYSIS: HOW DOES DEEP LEARNING UNVEIL NEW INSIGHTS IN ANCIENT TEXTS?
Author:	HASSAN HAZIMZE, SALMA GAOU, KHALID AKHLIL
Abstract:	Exploring the depths of history through its documents, the field of historical document analysis is critical in reconstructing our past. Traditionally, this field has relied on detailed manual examination by specialists. However, it is currently undergoing a significant transformation with the integration of deep learning methodologies, a shift that our study rigorously investigates. Emphasizing the transition from traditional techniques to deep learning approaches, particularly convolutional neural networks (CNN), we highlight how these advanced algorithms substantially automate and enhance the recognition and transcription of ancient scripts. This digital transformation not only expedites text processing but also achieves a level of precision surpassing that of manual methods. The core of our research is the innovative application of deep learning in character recognition, an essential step for accurately digitizing centuries-old manuscripts. We demonstrate the effectiveness of deep learning, especially CNNs, in identifying and converting diverse styles of handwritten script. This critical advancement is pivotal for preserving and thoroughly studying historical documents. Our findings reveal the profound impact of CNNs in enhancing both the accuracy and speed of character identification, marking a turning point in historical document analysis. By incorporating proven historical methodologies with advanced deep learning technologies, our study makes a substantial and explicit contribution to the field of historical document analysis. This fusion offers new perspectives for the detailed study of ancient texts and aids in their digital preservation. Our approach not only enriches our understanding of historical documents but also significantly enhances their analysis with the precision and accessibility afforded by advancements in deep learning. Conclusively, this research establishes a new paradigm in historical document analysis, addressing the research problem of efficient and accurate character recognition and contributing novel insights into the application of CNNs in this domain.
Keywords:	Character Recognition, Historical Documents, Deep Learning, Convolutional Neural Networks, Document Analysis, Information Extraction.
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	AN END-TO-END DEEP LEARNING APPROACH FOR AN INDIAN ENGLISH REPOSITORY
Author:	PRATHIBHA SUDHAKARAN, ASHWANI KUMAR YADAV, SUNIL KARAMCHANDANI
Abstract:	A voice recognition system that immediately translates raw audio waveforms into text without the need for separate components for language and acoustic modelling or manually constructed feature engineering is known as end-to-end deep learning for continuous speech recognition. It learns the whole audio to text mapping using a single deep neural network model. Conventional speech systems rely on intricate processing pipelines, but this method is far simpler. An end-to-end model in voice recognition is a simple single model that operates directly on words, subwords, or characters and may be trained from the ground up. This simplifies decoding by doing away with the requirement for both explicit phone modelling and a pronunciation lexicon. Deepspeech was used to construct and test the model, which was designed for Indian English. Additionally, a comparison is made between the results of the bi-directional RNN-based system and the traditional HMM model. With our method, we can quickly get a large amount of heterogeneous data for training due to a number of unique data synthesis techniques and an extremely effective RNN development system that utilises several GPUs. The connectionist temporal classification (CTC) objective function is used to infer the alignments between speech and label sequences, obviating the requirement for pre-generated frame labels. Experiments demonstrate that the RNN-based model has been observed to have equal word error rates (WERs) while also significantly speeding up the decoding process when compared to traditional hybrid HMM based on Kaldi.
Keywords:	End-To-End,HMM,CTC,RNN,CSR, Indian English, Deep Speech
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	RESEARCH TRENDS IN CONFIDENCE ASSESSMENT: A SYSTEMATIC LITERATURE REVIEW
Author:	HANA FAKHIRA ALMARZUKI, KHYRINA AIRIN FARIZA ABU SAMAH, LALA SEPTEM RIZA, SHARIFALILLAH NORDIN
Abstract:	Confidence assessment is vital for validating decisions, managing risks, and improving the quality of knowledge. However, there is a noticeable gap in comprehensive research aimed at investigating, understanding, and interpreting the evolving trends in confidence assessment. In light of this, the study involves conducting a systematic literature review to evaluate the nuances and key findings of previous works, shedding light on the current state of knowledge in this field. Emphasizing the significance of confidence, particularly in learning, is essential for accurately determining a student s level of knowledge. Our systematic review, following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, analyzes 39 studies (2018-2023) from Scopus, Web of Science, and Science Direct. Four primary themes—aims, methods, approaches, and Likert scale types are unfolded into 27 sub-themes, offering a comprehensive view of confidence assessment research trends. Notably, in fields like medicine, confidence assessment is pivotal, with pre-and post-surveys using a 5-point Likert scale being predominant. By synthesizing findings, it informs future research, enhances methods, and contributes to advancing our collective understanding of confidence assessment in knowledge creation.
Keywords:	Confidence Assessment, Systematic Literature Review, Knowledge, Confidence, Research Trend
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	BLOCKCHAIN BASED GENUINE AND TRANSPARENT CHARITY APPLICATION
Author:	ABHIJEET R. RAIPURKAR, MANOJ B. CHANDAK, AASHIA SORATHIA, ISHA MANKAR, PRASANNA TAPKIRE, PUSHKAR POPHALI
Abstract:	In an era where online philanthropy has gained momentum, ensured the authenticity of participants, and maintained transparency in transactions becomes imperative. Addressing these concerns, this research presents a two-fold solution. At the core of this system is a Decentralized Identity Verification feature, which ensures the authenticity of both donors and charity organizers. Participants validate their identities using established documents, which are then verified by respective issuing authorities. This platform enables users to initiate charity campaigns. Potential donors can view these campaigns, contribute as desired, and are further incentivized with a unique NFT (Non-Fungible Token) reward system upon their contributions. These NFTs, serving as digital memorabilia, commemorate the donor's charitable actions, adding an element of appreciation and potential digital value. Moreover, donors witness real-time transparent transactions, ensuring their contributions reach intended beneficiaries. Together, these systems, combined with the allure of NFTs, instill trust in digital charitable endeavors, paving the way for more secure, rewarding, and transparent online giving.
Keywords:	Blockchain, Decentralization, Smart Contracts, Rewards, NFT.
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	NOVEL DESIGN OF VOLUMETRIC ANTENNA FOR WIDEBAND APPLICATIONS
Author:	SURYA PRASADA RAO BORRA, A. GEETHA DEVI, , K. VIDYA SAGAR, SUMALATHA MADUGULA, MORSA CHAITANA, VUTUKURI SARVANI DUTI REKHA
Abstract:	This paper focused on designing a volumetric antenna to achieve high gain and high band width. For ground penetrating radars, medical microwave imaging to detect breast cancer, wideband antennas are significant. This paper focused on designing a volumetric wideband antenna to achieve high directivity high gain and high band width to address the challenges of both radar and medical imaging applications. The width of the antenna, effective dielectric constant, defective length, extension length, and micro strip patch length are meticulously tuned to achieve high directivity, high gain, and bandwidth. The proposed antenna is designed until the return losses, directivity, VSWR, gain and bandwidth values meet the requirements of volumetric wideband applications. The top layer is a hexagon patch. Octagonal is subtracted from hexagonal patch. And then circle is subtracted from the square. FR 4 is used as a dielectric substrate. Return loss (S11) parameter value is analysed at each stage of the antenna. The antenna Resonates at14.22GHz and15.33GHz, Voltage standing wave ratio (VSWR) is estimated, maximum peak gain (G) is achieved at 4.59dB. The band width achieved is 1.88. Input impedance is at 50Ω, current distribution and electric field distribution are evaluated. The VSWR and gain values, return losses, bandwidth are analysed at each stage of the design. The results achieved with the proposed antenna are matched with wide range of wireless sensor applications including Ku band and in specifically in medical era.
Keywords:	Bandwidth, Gain, Voltage Standing Wave Ratio (VSWR) And Volumetric Antenna, Return Losses.
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	BUILDING TRUST IN IOT: LEVERAGING CONSORTIUM BLOCKCHAIN FOR SECURE COMMUNICATIONS
Author:	MOHAMMED AMIN ALMAIAH, AITIZAZ ALI, RIMA SHISHAKLY, TAYSEER ALKHDOUR, ABDALWALI LUTFI, MAHMAOD ALRAWAD
Abstract:	As the Internet of Things (IoT) proliferates, ensuring the security and trustworthiness of communications becomes paramount. This paper introduces a novel approach to address these concerns by leveraging Consortium Blockchain technology. The proposed system focuses on building trust in IoT environments through a decentralized and transparent framework. We explore the integration of Consortium Blockchain as a foundational layer for secure communication within IoT ecosystems. The consortium model, involving a group of trusted entities, facilitates consensus mechanisms and smart contracts to establish and maintain a reliable reputation system. This approach mitigates traditional vulnerabilities associated with centralized systems and enhances the overall security posture of IoT networks. Key components of the proposed system include a consensus algorithm for agreement among consortium members, a transparent and immutable ledger for recording interactions, and smart contracts governing trust and reputation protocols. By utilizing blockchain technology, the system not only ensures data integrity and confidentiality but also instills confidence in the reliability of IoT devices and the information exchanged. Through simulation and analysis, we demonstrate the effectiveness of our Consortium Blockchain-based solution in enhancing the security and trustworthiness of IoT communications. The results indicate improved resistance to malicious attacks and a resilient foundation for building trust in the dynamic and interconnected world of IoT. This research contributes to the ongoing discourse on securing IoT ecosystems, offering a practical and scalable solution for building trust through Consortium Blockchain technology.
Keywords:	Internet of Things (IoT); Consortium Blockchain; Cyber-attacks; Trust in IoT.
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	OBJECT-ORIENTED ONLINE COURSE RECOMMENDATION SYSTEMS BASED ON DEEP NEURAL NETWORKS
Author:	HAO LUO, NOR AZURA HUSIN, SINA ABDIPOOR, TEH NORANIS MOHD ARIS, MOHD YUNUS SHARUM, MASLINA ZOLKEPLI
Abstract:	In the era of widespread online learning platforms, students commonly face the challenge of navigating an extensive array of available courses. Identifying relevant and fitting options aligned with students' educational objectives and interests is highly complex. The impact of system maintainability and scalability on escalated development costs is often neglected in the literature. To tackle these issues, this paper introduces a comprehensive analysis and design of an object-oriented online course recommendation system. Employing a deep neural network algorithm for course recommendation, our system adeptly captures user preferences, course attributes, and intricate relationships between them. This methodology facilitates the delivery of personalized course recommendations precisely tailored to individual needs and preferences. The incorporation of object-oriented design principles such as encapsulation, inheritance, and polymorphism ensure modularity, maintainability, and extensibility, thereby easing future system enhancements and adaptations. The main contribution of this paper is to propose a new idea of an adaptive learning system that combines deep learning for personalized recommendations with object-oriented design for scalability and continuous improvement. This practical solution demonstrably enhances online learning experiences by tailoring recommendations to individual needs and evolving trends. Evaluation of the proposed system's performance utilizes real-world online course datasets, demonstrating its efficacy in furnishing accurate and personalized course recommendations, ultimately enhancing the overall learning experience for students.
Keywords:	Object-oriented, Deep Neural Network, Online Course Recommendation, System Design
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	DEVELOPMENT AND SIMULATION OF MICROSTRIP PATCH ANTENNAS FOR 5G WIRELESS CONNECTIVITY
Author:	NEERU MALIK, SHRUTI VASHIST, AJAY N. PAITHANE, MUKIL. ALGIRISAMY
Abstract:	This study introduces a microstrip patch antenna designed to support 5G communication technology. It operates effectively at central frequencies of 38GHz and 54GHz, offering respective bandwidths of 1.94GHz and 2GHz. The antenna design prioritizes compactness, affordability, and suitability for miniature devices. It consists of an FR4 epoxy substrate, patch, and ground. The substrate boasts a dielectric constant of 3.8, a minimal loss tangent of 0.02, and adheres to a standard thickness of 1.57mm. The substrate measures 6mm x 6.25mm, and the patch's dimensions are 2mm x 2mm, employing the microstrip-line feeding technique.For mobile applications within the millimeter-wave spectrum, including frequencies of 38.6GHz, 47.7GHz, and 54.3GHz, accompanied by bandwidths of 3.5GHz, 2.5GHz, and 1.3GHz, this research also proposes an antenna array comprising four components, spaced at 4mm intervals. The overall antenna size is 6mm x 6.25mm x 0.578mm. The proposed antenna design undergoes rigorous simulation using HFSS software for performance validation.
Keywords:	5G, Microstripline,Antenna Array, Taperedline Feeding, HFSS Software.
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text

Title:	A NOVEL APPROACH TO UNIVERSAL SYSTEM WITH LOSSES THAT FOCUS ON THE ECONOMIC DISPATCH FOR THE ENERGY INTEGRATION IN MICRO GRID AND ACCENTUATE THE HEURISTIC OPTIMIZATION
Author:	V. SAI GEETHA LAKSHMI, M. VANITHASRI, M. VENU GOPAL RAO
Abstract:	Rapid development of renewable infrastructure and popular support for green energy have led to the emergence of hybrid generating systems in power networks that is micro grid. The effective scheduling of all power producing facilities to meet the increasing power demand is one of the most significant challenges in the design and operation of an electric power generation system. Scheduling power generating units to minimize costs and meet system restrictions is an example of economic load dispatch (ELD), a generic operation in the electrical power system. Due to their greater global solution capacity, flexibility, and deriv-ative-free construction, metaheuristic algorithms are rising in favor for addressing ELD problems. This study develops a novel hybrid optimization–based solution model for the ELD problem of integrating re-newable resources. In addition, this work takes into account multiple objectives, including the full cost of wind generation, the full cost function of thermal units, and a penalty cost function. The optimal output of thermal power plants is maximized using the hybrid optimization model. Limits, both upper and lower. In addition, the hybrid optimization model selects the best turbines to maximize wind energy production in response to specific needs. The efficiency and viability of the suggested algorithm were demonstrated using test system with 10 units. The Heuristic Optimization techniques is applied to get numerical results for the static and dynamic ELD problem show that the suggested elephant herding optimization (EHO) algorithm outperforms state-of-the-art methods in most of the test situations, proving its superiority and practicality.
Keywords:	Green Energy, Hybrid Generating Systems, Economic Load Dispatch, Heuristic Optimization, Elephant Herding Optimization.
Source:	Journal of Theoretical and Applied Information Technology 15^th February 2024 -- Vol. 101. No. 3-- 2024
Full Text