		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 in case of review papers).

Journal of Theoretical and Applied Information Technology
June 2024 | Vol. 102 No.11

Title:	DESIGN OF AN OPTIMIZED DEEP NETWORKS FOR FAKE QA INFORMATION PREDICTION SYSTEM
Author:	K. PUSHPA RANI, PELLAKURI VIDYULLATHA, Dr. K. SRINIVAS RAO
Abstract:	In recent years, the proliferation of fake information in Question-Answering (QA) systems has posed significant challenges for maintaining the integrity and trustworthiness of online platforms. To address this issue, we propose an optimized deep learning framework for fake QA information prediction, leveraging a novel combination of Squirrel Search Algorithm (SSA) and Extreme Gradient Boosting (XGBoost). Our approach, termed SSA-XGB, integrates the exploration and exploitation capabilities of SSA with the robust predictive power of XGBoost, resulting in an efficient and effective mechanism for detecting fraudulent content. The deep network architecture is meticulously designed to enhance feature extraction and representation learning, enabling it to discern subtle patterns indicative of fake information. Extensive experiments conducted on benchmark datasets demonstrate that SSA-XGB outperforms traditional machine learning and deep learning models with a recall of 96.7%, an F-measure of 96.6%, a precision of 99.6%, and an accuracy of 96.6%, while maintaining a low error rate of 4.0% and a computational time of 4.0 seconds. This innovative system offers a promising solution for safeguarding the quality of information in QA platforms, contributing to the broader effort of combating misinformation in digital ecosystems.
Keywords:	Topic Modelling, Relevant Answer, Squirrel Optimization, Gibbs Sampling, Features, QA System
Source:	Journal of Theoretical and Applied Information Technology 15^th June 2024 -- Vol. 102. No. 11-- 2024
Full Text

Title:	ENTERPRISE ARCHITECTURE FRAMEWORK IN THE GOVERNMENT SECTOR: A REVIEW
Author:	YOGA PRIHASTOMO, HARJANTO PRABOWO, AGUNG TRISETYARSO, HARYONO SOEPARNO
Abstract:	This study investigates the implementation landscape of enterprise architecture (EA) frameworks within the government sector. Despite the growing recognition of EA's importance, practical implementation in governmental bodies often needs help with challenges such as legacy systems, bureaucratic inertia, and resource constraints, leading to fragmented IT landscapes and inefficiencies. This study utilizes a systematic literature review (SLR) investigating the implementation landscape of EA frameworks within the government sector. Our analysis of diverse case studies across countries reveals notable trends in EA adoption, with Indonesia, Malaysia, and India emerging as leaders in implementation count. The prevalence of popular frameworks like TOGAF, Zachman, and FEAF is noted, alongside country-specific preferences such as Colombian GEAF, Namibian GEAF, Finnish National EA, and South Africa GWEA. Importantly, we identify six critical factors-Governance, Management, Resources, Socio-economic, Technology, and Information-that are crucial for successfully adopting EA frameworks in governmental contexts. This practical guide is designed to help policymakers and practitioners overcome implementation challenges, thereby enhancing organizational efficiency and governance.
Keywords:	Enterprise Architecture, EA Framework, e-Government, Systematic Literature Review, Critical Factors
Source:	Journal of Theoretical and Applied Information Technology 15^th June 2024 -- Vol. 102. No. 11-- 2024
Full Text

Title:	DETECTION OF CREDIT CARD FRAUD TRANSACTION USING HYBRID MACHINE LEARNING ALGORITHMS
Author:	BHUKYA DHARMA, DR.D. LATHA
Abstract:	In developed countries, credit card transactions are now the main method of payment, and their utility is growing quickly in developing. As a result, frauds are becoming a more serious issue, resulting in financial losses and a decline in customer trust. Firstly, the both real and fraudulent actors continually change their conduct, and secondly, is that datasets are wildly biased. There have been several suggestions for methods to handle the increasing number of credit card fraud transactions. To effectively identify fraudulent transactions, there has been use of machine learning techniques. This analysis explains the way to detect credit card fraud using a hybrid machine learning algorithm. The dataset utilized in September 2013 was the record of credit card transactions done by European cardholders over a duration of two days. Random Forest (RF) and Support Vector Machine (SVM) machine learning models are combined in hybrid categorization. The results of the Hybrid Machine Learning model are based on Accuracy, Sensitivity, Specificity, and Precision. Described model achieves Accuracy as 98%, Sensitivity as 96%, Specificity as 97%, and Precision as 96%. The outcomes from using the hybrid classification model have shown to be much more successful than those from using separate classification methods
Keywords:	Hybrid Machine Learning, Credit Card (CC), Fraud transactions, SVM, RF, Precision, Accuracy.
Source:	Journal of Theoretical and Applied Information Technology 15^th June 2024 -- Vol. 102. No. 11-- 2024
Full Text

Title:	TRAILBLAZING TECHNIQUES: EXPLOITING THE POWER OF TREES AND SWISH ACTIVATION FOR ENHANCED FEATURE RELEVANCE IN DEEP LEARNING
Author:	LEELAMBIKA KV, Dr. SHANMUGARATHINAM G
Abstract:	In recent years, the intersection of tree-based methods and advanced activation functions has led to remarkable advancements in deep learning. This paper presents novel trailblazing techniques that exploit the combined power of trees and Swish activation to significantly enhance feature relevance in deep learning models. By integrating feature importance derived from tree-based methods into feedforward neural networks (FNNs) with Swish activation functions, the proposed model, TreeFeatNet: Selective Feature Integration with FNNs, achieves superior performance in feature selection and model generalization. The effectiveness of the methodology was demonstrated through comprehensive experiments on diverse datasets across various domains. The results reveal that the synergy between tree-based feature importance and Swish activation facilitates the identification and utilization of highly relevant features, leading to improved model interpretability and predictive accuracy. Furthermore, the proposed techniques offer insights into the deep learning models, shedding light on the mechanisms underlying feature relevance and contributing to the advancement of interpretability in deep learning research. Overall, the study highlights the promising potential of integrating tree-based methods and Swish activation in deep learning, paving the way for future advancements in feature selection and model optimization.
Keywords:	Deep Learning, Tree-Based Methods, Swish Activation, Feature Importance, Neural Networks, TreeFeatNet: Selective Feature Integration with FNNs
Source:	Journal of Theoretical and Applied Information Technology 15^th June 2024 -- Vol. 102. No. 11-- 2024
Full Text

Title:	A NEW ALGORITHM FOR COMMUNITY DETECTION IN COMPLEX SOCIAL NETWORKS
Author:	HICHAM SADIKI, MEROUANE ERTEL, AZEDDINE SADQUI3, SAID AMALI
Abstract:	This article introduces a refined approach to identifying community structures in complex social networks. With a focus on accuracy and efficiency, our algorithm takes into account the complex nature of social networks by enhancing traditional methodologies to accurately capture community patterns. Central to our approach is the "community score," a pivotal metric gauging community partition quality. We've tailored variation operators, including a new crossover operator, to strengthen this foundation, improving both convergence and precision. A notable innovation is the dynamic determination of community count. Unlike fixed assumptions, our approach adapts the count based on network structure, adeptly detecting communities of diverse sizes and shapes. Moreover, we highlight border nodes' significance as community connectors. Weighted interactions involving these nodes improve community transition detection, refining partitions and spotlighting boundary-critical nodes. Through extensive experiments on synthetic and real-world datasets, the superiority of our algorithm over conventional methods becomes evident. Improved modularity and precision metrics validate our approach's efficacy.
Keywords:	Genetic Algorithm, Community Detection, Social Network Analysis, Graph Partitioning, Clustering, Complex Networks, Community Structure, Genetic Operators.
Source:	Journal of Theoretical and Applied Information Technology 15^th June 2024 -- Vol. 102. No. 11-- 2024
Full Text

Title:	THE ROLE OF VIRTUAL REALITY IN IMPROVING SOFTWARE TESTING METHODS AND TOOLS
Author:	IHOR HUNKO, OLEKSANDR MULIAREVYCH, RUSLAN TRISHCHUK, SERHII ZYBIN, ETAR HALACHEV
Abstract:	This report focused on the utilisation of VR in software testing methods and tools, highlighting its impact on immersive test environments, three-dimensional analysis, collaboration and remote testing, as well as automation and optimisation. Testing sites are based on VR solutions that turn virtual reality environment where testers can easily identify bugs and usability issues. Three-dimensional representations offer a look at the multiple functions of software and its interactive nature in virtual space; they also help in improving accuracy and diagnosis of test results. As the collaboration and remote testing can be carried out in VR suitable settings, it helps to overcome space barriers, ensuring seamless communication and team work between testers and developers. Compared with the traditional test processes, automation and optimisation in VR-based testing systemise the workloads, save labour and allocate the resources efficiently, therefore, improve the testing outcomes in quality and quantity. The paper carried out the advantages and difficulties that came with application of VR technology in software testing and later suggesting methods that could be employed when adopting VR-based testing methods. It not only shows different study areas where improvement of the software testing is possible but also helps to develop these areas by inspiring creativity, innovation and progress. Knowledge dissemination and conceptualisation strategies were proposed in the paper, and calls for collaboration as the approaches for bridging the knowledge gap and enabling a deeper appreciation of the VR-powered software testing potential. In the end, the main goal of this article was to provide stakeholders in software sphere with data about the ways a virtual reality technology can help to retrain the process of testing software, also to encourage organisations to find other ways for improving quality and user experience. By strategically incorporating VR technology, organisations can optimise their testing procedures to enhance the development of high-quality software products, ultimately strengthening their position in the rapidly evolving technological landscape.
Keywords:	Testing Software Programs, VR Technology, Debugging, VR Simulation, Manual Testing.
Source:	Journal of Theoretical and Applied Information Technology 15^th June 2024 -- Vol. 102. No. 11-- 2024
Full Text

Title:	ELEVATING PLASTIC LITTER SURVEILLANCE: DRONES AND DEEP LEARNING FOR EFFICIENT DETECTION
Author:	MIMOUN YANDOUZI, MOUNIR GRARI, MOHAMMED BERRAHAL, AISSA KERKOUR ELMIAD, MOHAMMED BADAOUI, ALAEEDDINE BARKAOUI, YASSINE ZARHLOULE
Abstract:	In our contemporary world, environmental issues and the ever-present threat of plastic pollution endanger not only the health of our planet but also that of its inhabitants, underscoring the urgency of action. The need to closely monitor these destructive phenomena and develop effective detection systems is imperative to preserve our fragile ecosystem. Fortunately, the emergence of cutting-edge technologies has revolutionized our ability to monitor and detect environmental threats with unprecedented precision. The combined use of drones and artificial intelligence, particularly deep learning, yields promising results, leveraging drones' unique capabilities to cover vast areas and the power of deep learning to analyze collected data swiftly and accurately. Our study focuses on optimizing the utilization of drones and object detection algorithms through deep learning for effective detection and supervision of plastic litter. We will explore the performance of two major families of object detection models, namely single-pass and double-pass, using drone images captured at varying heights. The overarching objective is to identify the optimal performance-to-resource conditions, maximizing efficiency in our detection and supervision endeavors. This research is crucial in addressing the pressing environmental concerns posed by plastic pollution, offering innovative solutions to mitigate its impact and safeguard the health of our planet for future generations.
Keywords:	Plastic Litter, Object Detection, Deep Learning, Drones/UAV, Faster R-CNN, YOLO
Source:	Journal of Theoretical and Applied Information Technology 15^th June 2024 -- Vol. 102. No. 11-- 2024
Full Text

Title:	HYBRID CASE RETRIEVAL USING FEATURE-VECTOR CASE REPRESENTATION IN A CBR E-LEARNING SYSTEM
Author:	ABROUN SOUNDOUSS, GHAILANI MOHAMED, FENNAN ABDELHADI
Abstract:	The COVID-19 pandemic has spurred the rapid adoption of e-learning systems, necessitating effective personalization strategies to cater to diverse learner needs. However, existing e-learning platforms often face challenges in providing tailored learning experiences. While existing Case-Based Reasoning (CBR) approaches in e-learning hold promise for personalization, their effectiveness hinges on robust case representation and retrieval methods. This paper addresses these limitations by proposing a hybrid case retrieval approach using feature-vector case representation for an Adaptive Intelligent Educational Distributed Case-Based Reasoning (AIED-CBR) system to capture comprehensive learner profiles This approach combines the strengths of rule-based and K-Nearest Neighbors (KNN) techniques within a Multi-Agent System architecture to enhance the efficiency and accuracy of case retrieval in personalized learning path generation. We leverage ontologies for knowledge description, facilitating efficient reasoning and knowledge sharing within the system. We present the system's architecture, detailing the hybrid retrieval mechanism and its integration with multi-agent collaboration, and the role of ontologies. This hybrid approach addresses limitations of existing CBR-based e-learning systems, offering the potential to create more effective and adaptable personalized learning experiences.
Keywords:	Case-Based Reasoning (CBR), Adaptive E-learning systems (AES), Ontologies, Multi-Agent System, K-Nearest Neighbors (KNN), Rule-based reasoning, Case representation.
Source:	Journal of Theoretical and Applied Information Technology 15^th June 2024 -- Vol. 102. No. 11-- 2024
Full Text

Title:	A METHOD TO MAKE THE ENCRYPTED COLOR IMAGE INCOMPREHENSIBLE AND USELESS
Author:	MOHAMMAD IBRAHIM AHMED AL-MAR, BELAL GHANEM MOHAMMAD AL-ATHAMNEH
Abstract:	The process of protecting a digital color image is an urgent necessity due to the confidentiality of the image or the possibility of it containing high-level data. In this research paper a new method of image cryptography will be introduces, tested and implemented. The proposed method will based on selecting a sequence of rotate left and exclusive operations. This sequence can be changed from time to time to increase the security level of image cryptography. The number of rotation digits can be changed; one or more private key can be used to ensure the image protection process. Some parameters such as MSE, PSNR and correlation coefficients will be calculated to show how this method will increase the distortion degree of the encrypted images, the results will compared with the XORing image cryptography to show the added damages to the encrypted image,
Keywords:	Cryptography, MSE, PSNR, Correlation Coefficient, Rotate Left, XORing, PK, Damage
Source:	Journal of Theoretical and Applied Information Technology 15^th June 2024 -- Vol. 102. No. 11-- 2024
Full Text

Title:	REVOLUTIONIZING COTTON FARMING: CO-CNN INTEGRATION FOR DISEASE IDENTIFICATION AND YIELD PREDICTION
Author:	S.GOVINDASAMY, D.JAYARAJ
Abstract:	In agriculture, accurate identification of cotton plant diseases and prediction of yield are crucial for ensuring crop health and optimizing production. This abstract explores the integration of Cassowary Optimization (CO) with Convolutional Neural Networks (CNNs) to enhance cotton plant disease identification and yield prediction. The CO-CNN framework demonstrates superior performance in accurately classifying instances and capturing underlying patterns in the data. By leveraging the dynamic optimization capabilities of CO, the model effectively optimizes CNN parameters, leading to improved convergence and performance. Results across various performance metrics, including Classification Accuracy, F-Measure, Fowlkes-Mallows Index, and Matthews Correlation Coefficient, showcase the efficacy of the CO-CNN model in addressing the complexities of real-world classification tasks. This innovative approach holds significant promise for empowering farmers and agronomists with advanced tools for early disease detection, yield prediction, and informed decision-making in crop management.
Keywords:	Cotton Plant Disease - Yield prediction – CNN – Cassowary Optimization – Disease Identification.
Source:	Journal of Theoretical and Applied Information Technology 15^th June 2024 -- Vol. 102. No. 11-- 2024
Full Text

Title:	A DETAILED CASE STUDY OF AN INTEGRATED REDUNDANT RELIABILITY MODEL USING THE PARALLEL-SERIES CONFIGURATION
Author:	BHAVANI KAPU, SRIDHAR AKIRI, SRINIVASA RAO VELAMPUDI, PAVAN KUMAR SUBBARA, RAMADEVI SURAPATI
Abstract:	The Integrated Redundant Reliability Model (IRRM) is a new approach to reliability engineering that strengthens system dependability by using a Parallel-Series Configuration. The efficiency of the system is higher than that of a single-system factor with an equivalent configuration, but the performance of each component within the parallel-series structure matters. The research provides an Integrated Reliability Model (IRM) that considers impacts in each phase, component efficiency, and current restrictions, specifically designed for the parallel-series scenario. Thanks to redundant components arranged in parallel inside subsystems, this architecture provides instantaneous backup for a single-phase AC synchronous generator. The interconnected series structure ensures operational continuity even in the event of a subsystem failure, reducing vulnerabilities associated with both parallel and series setups. The integrated approach's objective is to raise dependability levels; it is particularly useful for critical systems. The model uses Lagrangean methods to compute variable quantities, effectiveness, and phase dependability, and it considers several elements to increase overall system efficiency. Changes made to Newton-Raphson methodology and simulation techniques to ensure integer outputs add to the realism of the values collected. This research provides significant new understandings into how integrated redundancy strategies could optimize system dependability and efficiency.
Keywords:	IRRR Model, Lagrangean Approach, Component Reliability, Newton-Raphson Approach, System Reliability
Source:	Journal of Theoretical and Applied Information Technology 15^th June 2024 -- Vol. 102. No. 11-- 2024
Full Text

Title:	FUZZY DEM-SAW: A NOVEL HYBRIDIZED MODEL OF FUZZY DEMATEL-SAW IN LECTURERS’ PERFORMANCE EVALUATION BASED ON TEACHING, RESEARCH, SERVICE AND COMMERCIALIZATION (TRSC) CRITERIA FOR PROMOTION
Author:	PAUL AAZAGREYIR, JAMES AMI-NARH, DANIEL NII OKAIJAH WELBECK, WILLIAM LESLIE BROWN-ACQUAYE, ABA QUAYSON, JOEL OKOE QUARCOO, BISMARK DZAHENE-QUARSHIE
Abstract:	This study introduces an innovative hybridized model, Fuzzy DEM-SAW, designed to enhance the precision and efficacy of lecturers' performance evaluations for the purpose of promotion. This novel approach integrates two distinct methodologies, Fuzzy DEMATEL (Decision Making Trial and Evaluation Laboratory) and SAW (Simple Additive Weighting), presenting a comprehensive framework for the evaluation of lecturers based on the critical criteria of Teaching, Research, Service, and Commercialization (TRSC). The study’s purpose is to rank lecturers for promotion. The Fuzzy DEMATEL technique is employed to derive weights, serving as a fundamental basis for subsequent ranking through the Fuzzy SAW process. The proposed model is applied to a case study, revealing significant findings pertaining to lecturers' performance evaluation. The outcomes disclose that Benone secured the foremost position with an Si value of 0.7, followed by Begu-Ellah at 0.583, and Bemane at 0.488. These results provide valuable insights for decision-makers involved in the promotion evaluation process. This research not only contributes to the advancement of hybridized fuzzy models but also holds practical implications for optimizing the assessment of lecturers in academic institutions, thereby contributing to the broader discourse on performance evaluation methodologies in academia.
Keywords:	Fuzzy DEMATEL, Fuzzy SAW, Lecturers, Performance Evaluation, Artificial Intelligence
Source:	Journal of Theoretical and Applied Information Technology 15^th June 2024 -- Vol. 102. No. 11-- 2024
Full Text

Title:	FEATURES EXTRACTION USING GEOMETRY, STRUCTURAL, AND TEXTURE ANALYSIS FOR COVID-19 CLASSIFICATION FROM COMPUTED TOMOGRAPHY IMAGES OF THE LUNGS
Author:	MOHAMMAD ALFRAHEED
Abstract:	Worldwide publishing of the COVID-19 has occurred. One of its impacts has been demonstrated to be a lung infection in the patient. As the virus's virulence increased, the infection's dissemination throughout the lung increased. Utilizing computed tomography images (CT-Images), the severity of COVID-19 has been analyzed and diagnosed. Studies have consequently employed CT-Images to track patients' COVID-19 illness. The contribution here is to use CT-Images to extract the features of the infection. The proposed method uses several filters to eliminate shapes with insufficient grey intensity gradation before extracting the relevant features. The unique aspect of the method lies in the fact that different geometry, structural, and textural characteristics have been retrieved. The variety can be explained by having the ability to characterize the shapes of the infection utilizing those characteristics, in addition to preserving the internal and external appearance of the infection. The CT-Images have been classified using machine learning techniques, demonstrating the efficacy of these extracted features. The classification accuracy has varied between 99% and 100%, and many infection shapes have been identified in the CT-images. By increasing the classification accuracy, the proposed method has outperformed previous methods in terms of performance.
Keywords:	Computed Tomography Images, COVID-1, Feature Extraction, Classification, Machine Learning Algorithm.
Source:	Journal of Theoretical and Applied Information Technology 15^th June 2024 -- Vol. 102. No. 11-- 2024
Full Text

Title:	A HYBRID SARIMA-LSTM APPROACH FOR IMPROVED TIME SERIES PREDICTION OF AEROSOL OPTICAL DEPTH ACROSS DELHI,INDIA
Author:	NAUMI KRISHNA K. PANICKER, J. VALARMATHI
Abstract:	Atmospheric aerosols are one of the indispensable particles in understanding atmospheric dynamics and are essential for accurate environmental forecasting and policy development. The literature on AOD time series forecasting usually uses either statistical methods, which handle linear patterns but struggle with non-linearities, or machine learning (ML) and deep learning (DL) methods, which capture non-linearities but can be limited in accurately processing the linear components present in the data. This study introduces a hybrid model that combines statistical methods and ML techniques to effectively address both the linear and non-linear components present in AOD time series data. The primary goal of this work was to understand the potential of the hybrid SARIMA-LSTM (seasonal autoregressive integrated moving average—long short-term memory) model to enhance the forecasting capacity of AOD time series data. The proposed model was compared to its baseline models, SARIMA and LSTM, by utilizing monthly data from the Moderate Resolution Imaging Spectroradiometer (MODIS) satellite across the Delhi region of India from 2001 to 2019. The performance of these models was evaluated based on root mean square error (RMSE), coefficient of determination (R2), and mean absolute percentage error (MAPE) during both training and testing phases. The proposed model outperformed the baseline models in all three metrics. The findings of this study advocate hybrid modeling as a promising tool for improving the accuracy of time series prediction of AOD because it can handle both linear and non-linear aspects present in the data.
Keywords:	SARIMA, LSTM, Hybrid, Time Series Prediction, Forecasting, AOD, MODIS
Source:	Journal of Theoretical and Applied Information Technology 15^th June 2024 -- Vol. 102. No. 11-- 2024
Full Text

Title:	A DECISION SUPPORT MODEL TO IMPROVE COMPLAINT HANDLING IN E-COMMERCE TO ENHANCE CUSTOMER TRUST
Author:	YEHIA HELMY1, MERNA ASHRAF, AND LAILA ABDELHAMID
Abstract:	In today highly competitive environment, e-commerce businesses confront numerous challenges that threaten their sustainability. Trust is one of these challenges. Customer trust is the key element that ensures customer loyalty. Therefore, businesses striving to maintain a competitive advantage must make their customer the focal point of all operations. Effective complaint handling is one of the key elements for increasing customer trust. The complaint is an invaluable resource for businesses to retain customer trust and loyalty. Given the need to enhance the efficiency of addressing complaints from e-commerce customers, this study aims to revamp the complaint-handling process. This study proposes a new decision support model (E-CDSM) that integrates the automation concept. The model employs a classification approach to classify complaints according to their respective issues, clustering to group similar customers into batches, genetic algorithms to generate a list of suitable solutions for each batch, and a rule-based inference engine to produce instructions that aid staff in making optimal decisions for each complaint. The implementation of the E-CDSM shows a significant reduction in the processing time of complaints, with an increase in the accuracy of solutions provided to customers and the instructions provided to staff to make the best decisions. In turn, improved customer experience which resulted in an enhancement in customer trust and customer loyalty that retains business sustainability in the e-commerce market.
Keywords:	Customer Trust; Complaint Handling; Online Complaints; Text Classification; Decision Support Systems, and Business Sustainability.
Source:	Journal of Theoretical and Applied Information Technology 15^th June 2024 -- Vol. 102. No. 11-- 2024
Full Text