		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
July 2024 | Vol. 102 No.14

Title:	MAYFLY OPTIMIZATION WITH DEEP LEARNING ASSISTED GLAUCOMA DIAGNOSIS ON RETINAL FUNDUS IMAGES
Author:	ANITA MADONA M , DR. PANNEER AROKIARAJ S
Abstract:	Glaucoma is a chronic eye disease that causes vision impairment if not diagnosed and treated at earlier stages. Timely detection may save patients from permanent loss of vision. Physical examination of glaucoma by ophthalmologists contains costly, time-consuming, and skill-oriented processes. Various approaches are in investigational stage for identifying earlier-stage glaucoma, however, a sure diagnostic method remains challenging. Medical check-ups to observe the retinal area are occasionally required by ophthalmologists, who need a considerable number of experience and skill to appropriately interpret the outcomes. To overcome these issues, algorithm based on deep learning (DL) technique has been developed to examine imageries of optic nerve and retinal structures and to diagnose and screen glaucoma based on retinal input images. This article introduces a new Mayfly Optimization with Deep Learning Assisted Glaucoma Detection and Classification (MFODL-GDC) technique on Retinal Fundus Images. The MFODL-GDC technique aims to segment and categorize the retinal images for classification of Glaucoma. In the presented MFODL-GDC technique, bilateral filtering and CLAHE-based contrast enhancement are involved in image preprocessing. Besides, the MFODL-GDC technique applies Quick CapsNet model for optic disc (OD) and optic cup (OC) segmentation. Moreover, the MFODL-GDC technique uses DenseNet121 model for feature extraction and its hyperparameter tuning process can be performed by the use of the MFO algorithm. Furthermore, extreme learning machine (ELM) model can be exploited for the detection and classification process. The extensive performance validation of the MFODL-GDC technique is tested on benchmark datasets. The widespread comparison research stated that supremacy of MFODL-GDC technique over current techniques.
Keywords:	Glaucoma Screening; Computer-Aided Diagnosis; Retinal Fundus Imaging; Deep Learning; Mayfly Optimization
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	ENHANCED ROUTING PROTOCOL IN MOBILE AD HOC NETWORK FOR IMPROVING THE PERFORMANCE
Author:	DR.S.HEMALATHA, KOMALA C R, DR. KIRAN MAYEE ADAVALA, KHADRI SYED FAIZZ AHMAD, R.V.V. KRISHNA, B. DEEPA, DR.A.MOHAN, RADHA MOTHUKURI
Abstract:	Packet routing among the route path is a tedious task in Wireless Network and Creating an efficient route management system in wireless devices is a difficult task, particularly in Mobile Ad hoc Networks. Many studies are focused on offering efficient route management through the use of new algorithms and approaches. This article focuses on developing an upgraded routing protocol with route head support known as the Route Head based Routing Protocol (RHRP), which consists of three stages: route head forming, route prediction and packet forwarding. In the stage of route head , one node chosen and which takes the responsibility of packet forwarding , in route prediction stage the node decide the route path from the source node to the destination , and in the stage of packet forwarding flow the path to travel the packet to reach to the destination. The proposed RHRP was implemented in a network simulator and compared to existing routing protocols such as FLCH-AODV in terms of power analysis, end to end delay, energy consumption and connectivity analysis, the results shows that the proposed RHRP protocol is better. This proposed protocol also supports hidden and exposed node issues, buffer overflow, and energy optimization.
Keywords:	MANET, Network Layer, Route Head, Routing Protocol, Route Head Routing Protocol
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	IMPROVED SALP SWARM ALGORITHM FOR TEXT DOCUMENT CLUSTERING
Author:	WALEED ABDELKARIM ABUAIN
Abstract:	Text document clustering (TDC) is a crucial task in text mining that involves dividing a collection of documents into subgroups based on their level of similarity or dissimilarity. A vast amount of information is available on text clustering, and numerous attempts have been made to enhance the learning performance and address the TDC problem. One of the latest swarm algorithms based on population is the SALP Swarm Algorithm (SSA), which has been effectively applied to solve many optimization problems. However, the initial performance of SSA is limited to the exploitation phase, resulting in local optima trapping and a low convergence rate. This study proposes a novel approach to improve the SSA algorithm called the link-based SALP Swarm Algorithm (LBSSA), which enhances the exploitation capability of the original SSA. This involves adding an adjacent operator to the algorithm and utilizing a new aspect of probability, namely the neighborhood selection method (NSM), to improve the searching capability. The effectiveness of LBSSA was evaluated using six different text clustering datasets, demonstrating that the modified SSA combined with NSM significantly improved accuracy, precision, recall, F-measure, purity criterion, and convergence rate. Overall, LBSSA outperformed the original SSA algorithm and other popular clustering techniques such as K-means clustering, Density-based Spatial Clustering of Applications with Noise (DBSCAN), Agglomerative, and optimization algorithms such as Harmony Search (HS), Firefly Algorithm (FFA), BAT algorithm, particle swarm optimization (PSO), and Genetic Algorithm (GA).
Keywords:	Data Clustering, Test Clustering, Optimization, SALP Swarm Algorithm
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	DEVELOPING EFFECTIVE SOLUTIONS: RESEARCH DIRECTIONS AND IMPLEMENTATION STRATEGIES FOR EARLY RANSOMWARE DETECTION
Author:	ASMAA HATEM RASHID ABOGAMOUS
Abstract:	Ransomware attacks, employing advanced encryption to hold data hostage, pose a critical threat to targets ranging from individuals to critical infrastructure. Our study, analyzing over 150 references, reveals that 40% of research focuses on Detection Techniques, highlighting the urgency for early detection as traditional recovery methods falter once an attack begins. We review the progression of ransomware, attack methods, and detection datasets, offering a structured field overview. Attack Analysis and Patterns constitute 20% of the literature, followed by Prevention and Recovery 15%, and Cybersecurity Policy and Frameworks, as well as Evolution and Taxonomy of Ransomware nearly 10%. The remaining 5% covers surveys and comparative studies. Our findings underscore the need for improved ransomware detection capabilities and advocate for a multidisciplinary approach that combines technological innovation with an understanding of ransomware's development and classification to strengthen detection and prevention. This synthesis provides a snapshot of the current ransomware research landscape and underscores the imperative for ongoing investigation to counter these evolving cyber threats.
Keywords:	Ransomware Detection, Early Detection, Encryption Techniques, Cybersecurity, Detection Solutions, Ransomware Evolution.
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	ARTIFICIAL INTELLIGENCE-DRIVEN SOC PREDICTION IN ELECTRIC VEHICLES USING DBN-AOA
Author:	RAID MOHSEN ALHAZMI
Abstract:	Automobiles powered by electricity are an effective solution for the transportation sector's disastrous pollutant emissions. The performance of electric vehicles (EVs) is a determining factor in their massive and widespread acceptance among automotive consumers, despite the reality that their number of active users continues to rise. The EV industry has shown significant interest in lithium-ion batteries (LIBs) due to their cost-effectiveness, extended longevity, nominal voltage, and power density. State-of-charge (SOC) prediction accuracy is essential for effective battery management in EVs. However, non-linearities and complex dynamics inherent to LIBs pose challenges for traditional methods. This proposed work presents a novel deep-learning (DL) model for SOC prediction in EVs utilizing a Deep Belief Network (DBN) coupled with an Aquila optimization algorithm (AOA). The data utilized for training the proposed network is sourced from the SiCWell Dataset. The data is preprocessed through the implementation of Z-score normalization. The DBN utilizes battery data to extract and classify complex features, whereas the AOA is employed to optimize the hyperparameters of the DBN to increase the accuracy of predictions. The DBN+AOA is trained utilizing a SiCWell battery dataset in which the battery experienced a dynamic process. The performance of the DBN+AOA model is evaluated using the Mean Squared Error (MAE), Root Mean Squared Error (RMSE), and Mean Squared Error (MSE) metric values. Accurate results for SOC prediction are generated by the proposed method, with RMSE, MAE and MSE falling below 0.14%, 0.013%, and 0.011%, respectively. The average values of RMSE, MAE, and MSE are 0.136, 0.0122 and 0.0101. Experiments confirmed that the proposed DBN+AOA model has the best performance among the other current models in comparison.
Keywords:	EV, State-of-Charge, DBN, AOA, SiCWell, Lithium-Ion Batteries, Deep Learning.
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	TSUNAMI DISASTER POTENTIAL CLASSIFICATION USING C-LSTM METHOD
Author:	INDAH AYU KARUNIA WATI, ANTONI WIBOWO
Abstract:	Tsunami was one of the natural disasters caused by tectonic earthquakes, volcanic eruptions at sea, and underwater landslides. Classification or detection of tsunami disasters was very important to help detect tsunami disasters so that they could increase awareness and minimize the impact of losses. To perform classification, deep learning can be used. Therefore, this research aimed to classify the potential occurrence of tsunamis based on earthquake events that occured using the C-LSTM algorithm by configuring hyperparameter tuning implemented using the Python programming language. Hyperparameter tuning used such as filter, kernel_size, stride, optimizer, learning_rate, batch_size, num_lstm_layer. In this study a proportion of data 60%: 40%, 70%: 30%, 80%: 20%, 90%: 10% was used. Based on the results of the study, the highest accuracy was obtained at a proportion of 90%: 10% with the hyperparameters used, namely filter 64, kernel size 3, stride 1, the number of LSTM layers as many as 3 layers, using the adam optimizer with a learning rate of 0.001, and a batch size of 32 so as to obtain an accuracy of 91.46%, with a precision of 86.67%, recall of 89.66%, and f1-score of 88.14%.
Keywords:	C-LSTM, Classification, Tsunami, Hyperparameter Tuning
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	EFFICIENT DATA GATHERING MODEL WITH ENERGY BASED ROUTING FOR COMPRESSIVE SENSING IN MULTI-HOP HETEROGENEOUS WIRELESS SENSOR NETWORKS
Author:	BEJJAM KOMURAIAH, DR.M.S. ANURADHA
Abstract:	Lack of efficiency and infective routing are the major drawbacks in the compressive sensing based multi hope Heterogeneous Wireless Sensor Network (HWSN). To expand the longitivity and lifespan of the sensors it is very essential to concentrate on power utilization and routing models. For that purpose in this article Efficient Data Gathering Model with Energy based Routing is developed in compressive sensing based HWSN (EDGER-HWSN) network. The major categories of the proposed models are effective network model, radio and energy model, cluster formation algorithm, and inter cluster communication process. At the initial stage an effective network model is constructed which includes all the essential aspects of HWSN node construction with proper energy requirements. Followed by that a clustering process is performed that separates the normal sensors and the cluster heads (CH). The optimal CH is selected through this process and that’s pointers to reduce the latency and energy consumption among the heterogeneous nodes. With the presence of these models the efficiency of the network is improvised by the proper utilization of power among the sensor nodes and at each data transmission routing is also effectively monitored and the parameters like link failures and data loss are detected and neglected. The implementation of this model is performed in Network Simulation 2 (NS2) and the parameters analysis is performed concerned with nodes and sensors speed. At the end of the simulation it is proven that when compared with the earlier baseline methodologies EDGER-HWSN model performed better results concerned with power utilization, efficiency and throughput which enhances the functionalities of compressive sensing based HWSN network.
Keywords:	Heterogeneous Wireless Sensor Network (HWSN), Compressive Sensing, Cluster Formation Algorithm and Inter Cluster Communication Process
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	ADVANCED FUZZY DECISION SUPPORT MODEL FOR EVALUATING THE CRYPTOCURRENCY'S PERFORMANCE
Author:	MUHAMMAD FAHREZA DWINANTO, DITDIT NUGERAHA UTAMA
Abstract:	Cryptocurrency is a digital currency that uses a cryptographic system as a protection system so that transaction processes can be carried out safely. Digital currency is used as an investment which is seen as a trading commodity whose profits are derived from the difference between the buying price and the selling price. Of the many types of existing cryptocurrencies, an investor or layman who is new to cryptocurrencies must be careful in choosing investment options, because each cryptocurrency has performance (criteria). Therefore, to determine the performance of cryptocurrencies, an analysis is carried out using a Decision Support Model (DSM). DSM is used to evaluate the performance of cryptocurrencies in conducting transactions by applying the advanced fuzzy method. The data used comes from the Kaggle website "Cryptocurrency Historical Prices" and "Cryptocurrency Prices Data" for the period January 1 2021 to June 30 2022 which consists of 23 types of cryptocurrency. The parameters used are price trends, price consistency, and transaction volume consistency. Evaluation is done by performing verification and validation. The results of this study are cryptocurrency performance based on defuzzification using Mean of Maximum used to find the best type of cryptocurrency. The type of cryptocurrency that gets the best performance is Bitcoin with a performance value of 50.0. Meanwhile, the type of cryptocurrency that gets the worst performance value is Ethereum with a performance value of 29.5.
Keywords:	Advanced Fuzzy, Cryptocurrency, Decision Support Model, Evaluation, Mean of Maximum (MOM)
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	COMPARING WATCH DOG ALGORITHM AND CLASSIFICATION TECHNIQUES WITH MACHINE LEARNING TECHNIQUE TO DETECT THE VARIOUS ATTACKERS IN MOBILE ADHOC NETWORK
Author:	S. HEMALATHA, M TRUPTHI2, VANEETA M, DR NRIPENDRA NARAYAN DAS, R.V.V. KRISHNA5, DR. M. SURESH THANGAKRISHNAN, GAYATRI D. LONDHE, PONNURU ANUSHA
Abstract:	An Intruder and Attacker are everlasting problem in the packet communication. While facilitating communication among wireless nodes without relying on established infrastructure, networks become susceptible to security vulnerabilities. One such vulnerable network type is the Mobile Adhoc Network (MANET), where intruders and attackers play pivotal roles in compromising network integrity and performance. Various research endeavors focus on identifying and thwarting these threats, particularly targeting three types of attackers: black hole, white hole, and gray hole attackers, alongside intruders. This article delves into the implementation of the Watch Dog method, which monitors the forwarding times of each node in the communication process. Intruders are identified by delays in forwarding times, black hole attackers by dropped forwarded nodes, gray hole attackers by frequent delays in forwarded packets, and white hole attackers by nodes excessively forwarding packets to numerous recipients. Through the proposed Watch Dog Algorithm combined with Classification Techniques, implemented using network simulation, the efficacy of this approach is demonstrated. Comparative analysis against machine learning-based routing protocols reveals that the Watch Dog-based detection methods outperform, showing over 50% improvement, with performance metrics exceeding 90%.
Keywords:	Attackers, Black Hole Attackers, Gray Hole Attackers, Intruder, White Hole Attackers, MANET, Watch Dog Technique
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	A NEW TECHNIQUE FOR DETECTING EMAIL SPAM RISKS USING LSTM- PARTICLE SWARM OPTIMIZATION ALGORITHMS
Author:	TAYSEER ALKHDOUR, RANA ALRAWASHDEH, MOHAMMED ALMAIAH, ROMEL AL- ALI 4 SAID SALLOUM, THEYAZAN H.H ALDAHYANI
Abstract:	Unwanted email spam involves sending messages to numerous recipients, typically to market products, services, or scams without the recipient’s consent. These messages often contain false information. The goal of identifying email spam is to recognize and filter out undesired communications before they reach recipients' inboxes. Detecting spam emails is crucial for all involved parties, including users, companies, and email service providers. The detection of email spam impacts user satisfaction, security measures, trustworthiness, data security, network performance, cost management, adherence to regulations, reputation maintenance, industry norms, and the global email environment. By identifying and addressing email spam, individuals, businesses, and service providers can benefit from enhanced safety and effectiveness in the email network. The process of identifying email spam extends to email service providers, individuals, businesses, network managers, ISPs, security experts, regulatory bodies, data analysts, law enforcement agencies, cybersecurity entities, and developers of spam filtering software. Through the implementation of spam detection techniques, these entities can mitigate the risks associated with email spam and promote a secure and efficient email environment. In our methodology, we start by importing and preparing the data, followed by translating words into numerical sequences via word encoding. Subsequently, we train an LSTM network with a word embedding layer. We then select optimal solutions using the PSO algorithm and classify data using the trained LSTM network. Our results demonstrate that our approach enhances email spam detection and outperforms previous studies with metrics reaching up to 99.5%. We conclude that identifying email spam is essential for maintaining a smooth and reliable email platform. By detecting spam, users, companies, and email providers can improve user satisfaction, protect against cyber threats, conserve network resources, comply with regulations, and establish credibility with users.
Keywords:	Email spam; Cyber-Risks Cybersecurity attacks; LSTM; PSO; NLP.
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	ENHANCING WATER PURIFICATION EFFICIENCY THROUGH MACHINE LEARNING-DRIVEN MXENE FUNCTIONALIZATION
Author:	DR ARVINDER KOUR1, V VIDYASAGAR, DR. M. L. SURESH, DR. YOUSEF A. BAKER EL-EBIARY, RADHA MOTHUKURI, SWATI RAWAT, SHITAL P. DEHANKAR, DR. MOHAMMED SALEH AL ANSARI
Abstract:	Water purification is critical for sustaining life and safeguarding public health, yet existing methods face challenges such as low efficacy and high costs. This research examines the potential of MXENE materials to address these issues due to their unique structure and properties. The study aims to enhance MXENE functionalization techniques to maximize their effectiveness in water filtration through the application of machine learning approaches. By investigating various functionalization methods and leveraging machine learning to optimize MXENE characteristics, the research seeks to advance water purification technology. The novelty of this research lies in its integration of machine learning-driven methods for MXENE functionalization in water purification. By exploring novel approaches to modify MXENE characteristics and improve water filtration efficiency, the study contributes to addressing global water purification challenges. The study begins with an in-depth overview of MXENE substances, their synthesis techniques, and their relevance to water purification. It then delves into different functionalization procedures, emphasizing the importance of tailoring MXENE characteristics for specific water treatment applications. Machine learning approaches are proposed to forecast and optimize MXENE properties for enhanced water purification efficacy. The research demonstrates the potential of machine learning-driven MXENE functionalization in improving water purification processes. By optimizing MXENE characteristics, the efficacy of water filtration is significantly enhanced, addressing current limitations in purification technologies. The study concludes by highlighting the significance of its findings in addressing global water purification challenges. By overcoming obstacles through innovative approaches and leveraging machine learning techniques, the research underscores the potential impact of MXENE-based water purification methods in ensuring universal access to clean water.
Keywords:	Machine Learning, MXENE, Functionalization, Water Purification, Adsorption, Surface Chemistry.
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	AN EFFICIENT TRUSTED ROUTE WITH IDENTITY BASED NEIGHBOUR FEEDBACK LINKED CLUSTERING MODEL FOR SECURE DATA TRANSMISSION
Author:	ASWADHATI.SIRISHA, K.SANTHI SRI
Abstract:	Maintaining network security is essential for keeping data safe. It is the responsibility of system administrators to ensure that the functionality, usability, and security of a network are all adequately addressed. Access control, virus and anti-virus software, and other security measures can all be used to keep a computer network safe. Identifying malevolent sensor devices and eliminating the data they collect is crucial for mission-critical applications. Networks cannot directly use normal authentication and cryptography systems due to the limited resources of sensor devices. Consequently, to lessen the effect of malevolent sensors by efficient routing, an energy-efficient approach is required. The rapid growth in demand for network services and infrastructure in the last several decades has led to the global proliferation of static networks. The speed of the network's deployment is heavily dependent on the routing protocol chosen. The creation of a feasible and secure routing protocol is a must to meet the deployment needs while also satisfying the service level. Using hierarchical routing protocols based on Cluster Heads (CH) in combination with trust management strategies can be a useful option for creating a secure and reliable network where each node has complete trust in the next hop on its forwarding path. Using trust management concepts to develop a safe and attack-resistant protocol for routing in networks is as strong as ever. In this research, an efficient Trusted Route with Identity based Neighbour Feedback Linked Clustering (TR-INFLC) model is proposed for selecting the trusted route with linked clustering for secure data transmission. The proposed model when contrasted with the existing models, proposed model exhibits best performance.
Keywords:	Network Security, Trust Factor, Neighbour Feedback, Routing, Linked Clustering, Data Security.
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	PERFORMANCE COMPARISION OF APACHE SPARK AND SEQUENTIAL PROCESSING ON MACHINE LEARNING CLASSIFICATION ALGORITHM
Author:	N. SUDHAKAR YADAV, G. SURESH REDDY, A. SREENIVASA RAO, GANDLA SAI DHEERAJ RAO, MUKESH SAI PRODUTUR, P. NAYAB RASOOL KHAN, VALLAMKONDA ADARSH
Abstract:	Big data is a large collection of useful but often unstructured data. Machine learning uses this data to understand patterns and create models for analytical applications. Processing big data can be time-consuming, which is where frameworks like Apache Spark come in to help. These processing tools make real-time analytical applications more efficient and accurate. For example, credit card fraud detection uses big data frameworks to analyze transactions and predict whether they are fraudulent or valid based on certain attributes. This paper focuses on using Apache Spark for credit card fraud detection and compares its performance with sequential processing. The dataset used contains various features and over five lakh records labeled as fraud or valid transactions, stored in HDFS. The dataset is processed using the classification algorithm logistic regression in Spark's in-memory allotment, while the same dataset is processed sequentially and stored on the local system for comparison purposes. Performance comparisons are made based on metrics like RAM, CPU, network, disk usage monitored using Prometheus and Grafana monitoring tools. As the dataset size increases, Spark is expected to perform more efficiently compared to sequential processing. The user-defined implementation of logistic regression involves varying the threshold parameter value for equal sensitivity and specificity compared to the general threshold value which results in positive increases in accuracy, precision, sensitivity, specificity, and f1-score.
Keywords:	Big Data, Sequential Processing, Spark, Machine Learning, Logistic Regression
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	STRATEGIES FOR PROTECTING SENIOR CITIZENS AGAINST ONLINE BANKING FRAUD AND SCAMS: A SYSTEMATIC LITERATURE REVIEW
Author:	MUHAMMAD AMIRRUL ALHAFIZ BIN MOHD ZUKRY, MUHAMMAD NUR AQMAL BIN KHATIMAN, PROF. TS. RUSLI BIN HAJI ABDULLAH
Abstract:	This systematic literature review examines the need for strong strategies to protect seniors from online banking fraud and scams. The demographic's increased use of digital banking platforms due to the COVID-19 epidemic has increased their cyber risk. This study identifies and evaluates multifaceted strategies to improve digital literacy, create user-friendly digital banking interfaces, enact and enforce strict regulatory frameworks, and encourage senior citizens to use electronic banking post-COVID-19. Digital literacy empowers seniors by helping them navigate online banking platforms securely and spot scams. This requires operational proficiency, cybersecurity knowledge, and threat identification and response. Online banking platforms must be user-friendly. For seniors with various digital skills and physical limitations, straightforward and easy-to-use interfaces can reduce the risk of fraud. This comprises simplifying transaction processes, providing clear instructions, and providing customized support. Seniors using online banking need regulatory frameworks to protect their financial interests and privacy. This evaluation assesses the effectiveness of current fraud and scam laws and practices and the need for improvements to address senior folks' special vulnerabilities. It shows how technology and law interact, emphasizing that regulatory authorities must adapt to digital changes to ensure comprehensive protection. Seniors have adopted e-banking due to the COVID-19 epidemic, which forced a move to digital platforms for many daily activities, including banking. Seniors face a variety of online hazards, yet this transition offers convenience and accessibility. Trust-building, education, and support services are crucial to helping this generation adopt e-banking, according to the analysis. According to this analysis, older folks need a multi-pronged cybersecurity approach that includes technological, educational, and regulatory components to improve their online banking experience and safeguard them from fraud and scams.
Keywords:	Online Banking Security, Elderly Fraud Protection, Digital Literacy for Seniors, Cybersecurity Measures, User Interface Design, Regulatory Policies, E-Banking Adoption
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	A NODE AUDITOR BASED TRUSTED ROUTE SELECTION WITH NODE AUTHENTICATION USING MULTI KEY DISTRIBUTION MODEL FOR SECURE DATA TRANSMISSION
Author:	ORCHU ARUNA, MIDHUNCHAKKARAVARTHY
Abstract:	Wireless Sensor Network (WSN) is an advanced and difficult-to-implement system that makes minimal use of computer resources. Security is a major concern in the WSN. It is susceptible to attacks and data packet loss because to its wireless nature. Avoiding such issues necessitates the use of secure routing. Routing is a crucial WSN technique for ensuring the safety of the network by distributing data to other nodes. The predicted trust value is used by the routing process's trust algorithm to either exclude or include nodes. Multiple secure communication protocols have been developed and deployed in WSNs to guarantee the confidentiality, integrity, and availability of the data and nodes involved. The importance of trusted communication in WSNs applications is growing in order to ensure their widespread adoption. A WSN platform needs a TMS in order to set up a trustworthy connection. This study suggests a different way to do secure sensor communication, building on ideas put forward by the secure Computing Group. This model suggests a secure routing protocol that takes trust into account in order to safeguard wireless sensor networks from different types of threats. Using a cryptographic architecture for node authentication, this study chooses an Auditor node to carry out route selection in the WSN. To prevent attacks such as black holes, selective forwarding, wormholes, hello floods, and sinkholes, nodes must first calculate the overall trust values of their neighbors by combining the direct and indirect trust values, as well as the volatilizing factor and residual energy. Sending a routing request message to neighbors in multi-path mode is the initial step for a source node to transfer data to a sink node. The batteries used to power WSN nodes have a very short lifespan of only a few days. As a rule, sensor nodes are deployed in convoluted places, making it difficult to access them for battery maintenance or recharging. As a result, employing the elaborate procedure for securing data is not recommended. In this research, Node Auditor based Trusted Route Selection with Node Authentication using Multi Key Distribution (NAbTRS-NA-MKD) Model is proposed for secure data transmission in WSN. The proposed model when compared with the traditional methods exhibits 98.8% accuracy in trust factor calculation and trusted route selection.
Keywords:	Wireless Sensor Networks, Routing, Auditor Node, Trusted Node, Node Authentication, Multikey Distribution, Secure Data Transmission.
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	A NOVEL APPROACH TO MODELING TOPICS THROUGH A DISTRIBUTED FILE SYSTEM FOR JOB ASSISTANCE IN SOCIETAL COMMUNICATION EMPOWERMENT
Author:	K. PUSHPA RANI, PELLAKURI VIDYULLATHA, Dr. K. SRINIVAS RAO
Abstract:	The current research program uses data from social networks to explore public opinion on technical terms or topics. For job seekers, understanding how the general public perceives technical phrases or subjects and their impact on the environment and society is crucial. Public support is also critical for legislation and the implementation of mitigation programs. Public opinion research is essential for a better understanding of the social environment and dynamics. Social media data provides valuable information on public attitudes and responses to conflicting socio-technical terms or issues from various perspectives, such as quorum, stack overflow, and Yahoo!. It responds to Twitter, among other platforms, and is frequently used to track and assess how society responds to a natural or societal anomaly. Typically, social media data is acquired by searching for keywords or a specific topic to identify various topics in the topic templates. However, in conventional topic models, users can provide an inaccurate number of topics, leading to subpar grouping outcomes. Accurate representations are crucial for retrieving data and identifying cluster trends. To address this issue, viable methods for modeling themes are related to unclassified and incorrect texts or topics. The Distributed Latent Semantic Analysis (DLSA) and the Distributed Latent Dirichlet Allocation (dLDA) are two techniques used for this purpose.This document provides a brief overview of the country's public question-and-answer system and traces the evolution of significant issues and initiatives, paying particular attention to the automatic dissemination of pertinent customer feedback and knowledge of relevant awareness-raising information. It also highlights opportunities for housing and employment for the newest technologies in global empowerment. Finally, the experimental findings suggest that topic models outperform existing models in terms of precision for obtaining more pertinent responses from a placement and interview perspective.The research addresses the challenge of accurately modeling themes in social media data to understand public opinion on technical terms and topics. By employing advanced techniques such as DLSA and dLDA, the study enhances the precision of topic modeling, leading to better data retrieval and identification of cluster trends. This improvement aids job seekers in understanding public perception, supports legislative efforts, and facilitates the implementation of mitigation programs. The impact of this research lies in its contribution to more effective public opinion analysis, thereby informing policy-making and societal responses to technical and environmental issues.
Keywords:	F-Score, Hadoop, LSA, LDA, overflow, Quora, Topic models, stack, Twitter API
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	FEATURE EXTRACTION OF EEG SIGNALS IN THE TIME-FREQUENCY DOMAIN OF REHABILITATION TASK FOR MOTOR IMAGERY BRAIN-COMPUTER INTERFACE IN UPPER LIMBS
Author:	DAINIER GONZÁLEZ ROMERO, NUÑEZ CUELLAR SEBASTIAN, NIETO VALENZUELA SERGIO ALEJANDRO, RUTHBER RODRÍGUEZ SERREZUELA, ENRIQUE MARAÑÓN REYES, ARQUÍMEDES MONTOYA PEDRÓN, ROBERTO SAGARÓ ZAMORA
Abstract:	Effective feature extraction and classification methods of electroencephalogram (EEG) signals are critical for enhancing the recognition accuracy of Brain-Computer Interface (BCI) systems used in disability assistive devices and rehabilitation equipment. This study aims to identify and integrate EEG signals that classify real and motor imagination (IM) movements within a BCI system for use in robotic technology in post-stroke patient rehabilitation. We propose a processing method that combines low and high pass filtering, principal component analysis (PCA), and time-frequency domain signal processing using the Fourier Transform. Signals were recorded from healthy subjects and synchronized with recording software via a custom interface, focusing on alpha and beta brain rhythms. After filtering, PCA was used to reduce the number of reading channels for each rhythm. A primary challenge addressed is the inconsistency in registering real and motor imagination movements based on EEG signals. Our methodology includes developing a visual interface using Matlab 2019 and the Neuronic Cognitive Stimulator to guide users through movement tasks while synchronizing and recording EEG data. Bandpass filters for alpha, mu, and beta rhythms were designed, with the Butterworth filter chosen for its optimal balance of performance and computational cost. PCA identified the most relevant EEG channels, reducing data dimensionality while preserving critical information. Fourier and Fast Fourier Transforms (FFT) were applied to differentiate movements based on frequency analysis. This approach successfully identified a fundamental frequency one second before movement execution, facilitating the recognition of movement intention. The developed system shows promise for improving the precision of BCI applications in neurorehabilitation, providing a flexible tool that can incorporate other signal extraction methods.
Keywords:	Feature extraction, principal components analysis, EEG signals.
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	MODELLING AN EFFICIENT REGULARIZED OPTIMIZATION APPROACH FOR HEALTH CARE APPLICATION
Author:	K.GUNASEKARAN, V.D.AMBETH KUMAR, P. SHERUBHA, S.P. SASIREKHA
Abstract:	Chronic conditions like diabetes, heart disease, cancer, and chronic respiratory disorders pose a threat to people everywhere. Among these, the diagnosis of heart disease is made more difficult by its variable symptoms or traits. Internets of Things (IoT) solutions are crucial for healthcare detection. The suggested approach combines fog, edge and cloud computing to deliver quick and accurate results. The hardware elements gather information from various patients. To obtain important features, signals are subjected to cardiac feature extraction. Additionally, data on the feature extraction of other properties are acquired. An optimized cascaded convolution neural network collects all these features and subjects them to the detection system. Squirrel Optimizer is adopted over the auto-encoder (SOAE) technique to optimize the AE hyper-parameters. The suggested SOAE is 4%, 4%, 4%, 8%, 68%, 49%, 34%, 11% and 8% more accurate than PSO, GWO, WOA, DHO, DNN, RNN, LSTM, CNN, and RCNN, respectively, according to performance studies. The comparison analysis shows that the proposed system performs better than conventional models.
Keywords:	Regularized Model, Encoding, Optimization, Health Care, Prediction
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	RELATED FEATURE SUBSET MODEL FOR CREDIT CARD FRAUD DETECTION USING MACHINE LEARNING METHODS
Author:	KALPANA PAWASE, VAIBHAV VASUDEVARAO GIJARE, NAGAMANI CHIPPADA, JAYAKRISHNA AMATHI ,SANJAY CHABILDAS PATIL
Abstract:	The Internet has experienced exponential growth over the past decade. Subsequently, the prevalence and prominence of services such as e-commerce, swipe and pay, and online bill payment have increased. Subsequently, criminals have intensified their endeavors to compromise credit card transactions. In the event that consumers are billed for items they did not purchase, it is imperative for credit card companies to possess the capability to identify fraudulent transactions. Data Science and Machine Learning are indispensable for resolving problems of this nature; their significance cannot be overstated. An increasing number of customers are demanding more amenities from businesses. An instance of such convenience is the capability of conducting online product purchases. The objective of this study is to illustrate the application of machine learning in the construction of a credit card fraud detection dataset. The credit card fraud detection problem involves the incorporation of data from successful credit card transactions into models of previous transactions. It is possible to ascertain the legitimacy of a new transaction by employing these methods. The prevalence of credit card fraud has increased in tandem with advancements in electronic payment systems and e-commerce. Procedures for detecting credit card fraud must therefore be implemented. When employing machine learning techniques for credit card fraud detection, it is vital to exercise extreme caution when selecting the characteristics of fraudulent transactions. This research presents a Related Feature Subset Model for Credit card Fraud Detection (RFSM-CFD) for accurate detection of credit card frauds. Feature selection for the machine learning -based credit card fraud detection system is proposed in this research. This research achieves 98.8% accuracy in feature subset generation and 98.5% accuracy in credit card fraud detection. When compared to state-of-the-art models, the proposed fraud detection model demonstrates superior accuracy. The outcomes indicate that the proposed model outperforms conventional models.
Keywords:	Credit Card, Fraud Detection, Machine Learning, Feature Set, Subset Model, Transaction Details, Credentials, Attackers.
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	OPTIMIZED ENSEMBLE LEARNING FOR SOFTWARE DEFECT PREDIC-TION WITH HYPERPARAMETER TUNING
Author:	GETACHEW MEKURIA HABTEMARIAM, SUDHIR KUMAR MOHAPATRA, HUSSIEN WORKU SEID
Abstract:	The Software plays a crucial role in human life, making it essential for system developers to have reliable and accurate software. The discovery of faults during software development is becoming increasingly important to minimize costs and delivery time. As the application of software in business increases then, the soundness of software becomes more important. Several logical models have been proposed to evaluate software system reliability and predict software trustworthiness but the existing reliability model may be efficient towards solving a specific type of problem but incapable of solving other classes of software problems. Therefore, a novel and universal model is needed for fair prediction and error classification of all types of software reliability prediction problems. The study proposed ensemble models for software reliability prediction, which have an advantage over existing statistical and machine learning models. The proposed model is a binary model that can be used for error classification with automatic hyperparameter selection for flexibility. A total of 21 static metrics of the NASA dataset are taken as independent variables for the classification model and bagging, voting and stacking techniques are applied for classification. The performance of the models was evaluated using accuracy, precision; recall and f1-score and the model achieved 89.1% classification accuracy. The proposed ensemble model was also compared with existing models using a benchmark dataset for their performance. The results of the statistical comparison of the proposed model show better performance as compared to other existing models.
Keywords:	Software reliability, Software reliability prediction, Software reliability classification, Ensemble model, Machine learning, Hyperparameter
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	IDENTIFYING COVID’19 SEVERITY USING DEEP LEARNING MODELS FROM CHEST X-RAYS
Author:	M. COUMARANE, M. BALASUBRAMANIAN , S. SATHIYA
Abstract:	TThe contribution of Artificial Intelligence to the medical field in diagnosing various ailments have been triggered in the recent years and one such impact is the exposure of corona virus in the most reliabl emanner. Recently the familiar infection COVID 19 caused by coronavirus was first discovered in Wuhan in the end of the year which became major disaster for this century. The disease spread rapidly without control and the only remedy was preventive measures taken in advance. The examination depended on chest x-ray images which were processed to differentiate from pneumonia and other cold related diseases. This paper is related to the images identified as corona virus and the sternness of the disease is categorized into three classes. The images are segmented and classified using various deep learning techniques like Residual network, Exception model and Dense net model for comparison purposes. The best model is chosen from the accuracy produced with the given data sets. The deep learning Xception model proves to be the best with overall accuracy of 89% in identifying the disease using the chest x-ray images
Keywords:	Artificial Intelligence, corona virus, COVID 19, X-ray images, pneumonia, deep learning techniques, Residual network, Exception model, Dense net model
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	IMPROVING QUALITY OF SERVICE IN MOBILE ADHOC NETWORK BY DOING MISSING PACKET COLLECTION DUE TO BUFFER OVERFLOW WITH DIVIDE AND CONQUER STRATEGY
Author:	S.HEMALATHA, GAYATHRI G N, T.THILAGAM3, DR.SUNITA RAWAT, R.V.V. KRISHNA, PRAMODKUMAR H KULKARNI, DR.A.MOHAN, RADHA MOTHUKURI
Abstract:	One of the difficult problems in Mobile Adhoc Network is enhancing the Quality of Service which can be accomplished through effective packet delivery between source and destination nodes. An improved buffer management method that recovers packet loss due to buffer overflow results in a higher packet delivery ratio. Many research works were conducted to overcome packet loss, but all required some additional overload to the routing algorithm, transport layer, and could not achieve the expected results. Furthermore, existing researches could not concentrate a new vision about the collection of missing packets that could support QoS in the MANET. This research article goal was discovered to be achieving the QoS of the MANET by facilitating packet delivery to the destination. Furthermore, rather than resending all packets to the destination, the research concentrated on missing packets that were dropped due to congestion or buffer overflow in the node, as well as collecting those missing packets from the intermediate node in the route from the source to the destination. This article focuses on locating and forwarding lost packets utilizing the divide and conquer method of the route path. The proposed Divide and Conquer AODV (DVCAODV) buffer management was simulated with NS2.34 and compared with the available buffer management PBMTAODV, TCP/IPAODV, and HDELL-MCTOADV, with simulation values revealing that packet loss ratio and missing packet collection time of the proposed DVCAODV ranged from 0.04% to 0.09%.and 0.07ns to 0.12 ns respectively.
Keywords:	Buffer management, Congestion control, Divide and Conquer, Mobile Ad Hoc Networks (MANETs), Packet delivery, Quality of Service (QoS).
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	IMPROVING SECURE ROUTING IN IOMT: ENHANCED BLOCKCHAIN CYBERSECURITY SCHEME USING HYPERLEDGER FABRIC
Author:	TAYSEER ALKHDOUR, MOHAMMED AMIN ALMAIAH, AITIZAZ ALI, ROMEL AL-ALI4, TING TIN TIN, THEYAZAN ALDAHYANI
Abstract:	The Internet of Medical Things (IOMT) has emerged as a transformative technology in the healthcare sector, enabling seamless monitoring and management of patients’ vital health data. However, the integration of IOMT into healthcare ecosystems raises critical concerns about data security, especially during data transmission and routing. In this context, we present a novel approach for secure routing in IOMT by combining the power of homomorphic encryption and permissioned blockchain technology using Hyperledger Fabric. Our proposed framework addresses the pressing need for confidentiality, integrity, and authenticity of medical data as it traverses through interconnected IoT devices and networks. To achieve this, we leverage homomorphic encryption to perform computations on encrypted data without decrypting it, preserving patient privacy while enabling data analysis. Furthermore, we introduce a permissioned blockchain network built on Hyperledger Fabric to establish a trust infrastructure among healthcare entities, ensuring that only authorized nodes can participate in the routing process. Through the integration of homomorphic encryption and Hyperledger Fabric, our approach guarantees end-to-end security during data routing in IOMT. We discuss the architecture, components, and protocols that facilitate secure routing and present a comprehensive evaluation of the framework’s performance and security properties. Our results demonstrate the efficacy of this approach in safeguarding sensitive medical data and preserving patient confidentiality, opening up new possibilities for secure and privacy-preserving IOMT applications in healthcare. This research contributes to the ongoing efforts to enhance the security of IOMT systems, addressing a critical concern in the adoption of these technologies within healthcare and related domains. The fusion of homomorphic encryption and permissioned blockchain not only fortifies data routing security but also lays the foundation for the development of resilient and trust-based healthcare ecosystems in the era of the Internet of Medical Things.
Keywords:	Internet of Medical Things (IOMT); Blockchain; Cybersecurity and Hyperledger Fabric.
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	AI-DRIVEN CHATBOT IMPLEMENTATION FOR ENHANCING CUSTOMER SERVICE IN HIGHER EDUCATION: A CASE STUDY FROM UNIVERSITAS NEGERI SEMARANG
Author:	MUHAMAD ANBIYA NUR ISLAM, BUDI WARSITO, OKY DWI NURHAYATI
Abstract:	Given the limited human resources and the needs of service users at Universitas Negeri Semarang (UNNES) helpdesk, there is a need for a solution regarding service problems. This study aimed to implement and evaluate an integrated chatbot system using similarity-based and generative-based response generation models at UNNES' helpdesk. The primary contribution is enhancing response efficiency and user satisfaction through automated, context-aware responses, which is a novel approach in higher education institutions. The primary objective was to enhance response efficiency and user satisfaction using automated and context-aware response generation. The methodology involved deploying the TF-IDF model for initial query handling to quickly retrieve relevant Frequently Asked Questions (FAQ) responses. Additionally, a generative model, Llama RAG, was employed for generating nuanced answers when queries fell below a defined similarity threshold. The steps included data collection, preprocessing, model training, and performance evaluation using precision, recall, F1 score, and BLEU score metrics. The TF-IDF model effectively handled 78% of queries, while the Llama RAG model addressed the remaining 22%. The average similarity score of TF-IDF responses was 0.85, and the BLEU score for generative responses was 0.61, demonstrating high relevance and linguistic coherence, respectively. These findings underscore the potential of integrating advanced AI models to improve helpdesk operations, suggesting that such systems can significantly enhance user interaction and operational efficiency.
Keywords:	AI-Driven Chatbots, Customer Service Automation, Natural Language Processing, Higher Education Helpdesk, Hybrid Chatbot Systems
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	OPTIMIZING SOLAR PV SYSTEM FOR SECOND- ORDER FUZZY LOGIC INVERTER DESIGN FOR UPQC TO ENHANCE POWER QUALITY
Author:	G.SUJATHA, S.VENKATA PADMAVATHI
Abstract:	Solar photovoltaic (PV) technology uses photovoltaic effect, solar cells convert sunlight directly into electrical energy, offering a clean and sustainable power source. With advancements in technology and decreasing costs, solar PV has emerged as a cost-effective and environmentally friendly solution for generating electricity, playing a vital role in the transition towards a more sustainable energy future. The Unified Power Quality Conditioner (UPQC) plays a crucial role in addressing power quality issues associated with PV systems. It effectively mitigates voltage sags, swells, fluctuations, and harmonic distortions caused by the intermittent nature of solar energy generation. This paper presented an Enhanced Second-Order Generalized Integrator (ESOGI) control strategy for the PV application. The proposed ESOGI model uses the Fuzzy logic scheme with the Second -Order Generalized Integrator (ESOGI) model for the PV. With the uses of the ESOGI model second-order-based fuzzy logic model for the estimation of load in the PV for the different variations of load in the applications. The ESOGI model utilizes quasi-z-source inverter for the PV application for the UPQC model for PV. This paper investigates the implementation of the Enhanced Second-Order Generalized Integrator (ESOGI) control strategy for the Unified Power Quality Conditioner (UPQC) in Solar Photovoltaic (PV) applications. Through a comparative analysis with conventional control techniques such as PI Control, PID Control, and PWM Control, the efficacy of ESOGI is evaluated across various parameters including Total Harmonic Distortion (THD), voltage regulation, power factor improvement, and reactive and real power compensation. The ESOGI control strategy offers enhanced capabilities in improving power quality, fault-ride-through performance, and system stability during transient conditions. Through a comparative analysis with conventional control techniques such as PI Control, PID Control, and PWM Control, the efficacy of ESOGI is evaluated across various parameters including Total Harmonic Distortion (THD), voltage regulation, power factor improvement, and reactive and real power compensation.
Keywords:	Solar Photovoltaic (SPV), Unified Power Quality Conditioner (UPQC), Fuzzy Logic.
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text

Title:	A NEW IMAGE DENOISING TECHNIQUE VIA MACHINE LEARNING
Author:	RASYIDA MD SAAD , WAN ZUKI AZMAN WAN MUHAMAD , ACHMAD ABDURRAZZAQ
Abstract:	Image denoising is a pre-processing process usually performed to remove all the noises that might hinder the process of extracting information from an image. Hence several methods have been proposed to tackle this problem, especially for salt and pepper noise. However, the result of the recovery image especially the noisy image with higher noise densities is unsatisfied either the noises are not cleaned or the blurry effect is noticeable on the recovery image. In this paper, a new technique is proposed by infusing the use of a machine learning technique which is a support vector machine (SVM), and an existing image filtering method which is a median filter. Acknowledging that the use of a median filter alone will affect the edge of the recovery image including a blurry effect, the apparent change in the result either qualitative or quantitative can be seen when combining the use of a median filter with the SVM. 8 grayscale images contaminated with salt and pepper noise are used to validate the proposed technique. A comparison with the existing methods in terms of image quality assessment tests is also performed to validate the effectiveness of this technique and from the result of recovery images, it can be seen that the proposed technique had shown a favorable result in terms of qualitative and quantitative results as compared to the other existing methods.
Keywords:	Image Denoising, Machine Learning, Support Vector Machine, Singular Value Decomposition, Median Filter
Source:	Journal of Theoretical and Applied Information Technology 31^st July 2024 -- Vol. 101. No. 14-- 2024
Full Text