		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
March 2024 | Vol. 102 No.5

Title:	THE IMPROVEMENT PREDICTION MODEL USING ANFIS FOR MEDICAL DATASET
Author:	SRI SUMARLINDA, AZIZAH BINTI RAHMAT, ZALIZAH BINTI AWANG LONG, WIJI LESTARI
Abstract:	The prediction model developing with better performance can be used for early detection of heart disease and stroke for first step healthy care. Improving the performance of this prediction model is related to solving problems in terms of convergence, overfitting and underfitting. This research aims to develop a prediction model using ANFIS (Adaptive Neuro-Fuzzy Inference System) to detect early heart disease and stroke. The dataset used consists of 500 data with 12 features, covering various risk factors such as blood sugar levels (blood sugar), cholesterol, uric acid, systolic blood pressure, diastolic blood pressure, body mass index (BMI), age, smoking habits, lifestyle, genetic factors and gender and 1 label feature. The prediction model with ANFIS is implemented in three different models with varying learning rates to increase accuracy and prediction performance. In this research, Model 1 used a percentage of 60% training data and 40% testing data. Model 2 used a percentage of 70% training data and 30% testing data, while Model 3 used a percentage of 80% training data and 20% testing data. All three models show good accuracy and performance, namely above 89%. Model 2 has an accuracy value for training data of 0.980000, while for testing data it is 0.913333, showing the best performance compared to Models 1 and 3. Furthermore, learning rate variations were carried out on Model 2 with values of 0.01, 0.05, 0.1, 0.2, and 0.5. The best prediction process was obtained at a learning rate of 0.1. The Root Mean Square Error (RMSE) value for the training data is 0.050727, with an accuracy value of 0.985714, and an F1-Score value of 0.990253. Meanwhile, for testing data, the RMSE is 0.537474, with an accuracy value of 0.900000, and an F1-Score value of 0.928910. Thus, it can be concluded that the best model in this research is Model 2 with a learning rate of 0.1.
Keywords:	Improvement Prediction Model, ANFIS, Dataset
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	MEASUREMENT OF SEMANTIC TEXT SIMILARITY
Author:	ZHENGFANG HE, CRISTINA E. DUMDUMAYA, VAL A. QUIMNO
Abstract:	Semantic text similarity measurement is fundamental in natural language processing (NLP). With the advancement of NLP technology, the research and application values of similarity measurement have become prominent. This paper utilizes Google Scholar as the primary search tool and collects 179 documents. Then, using filtering technology, 50 key documents are ultimately obtained. Furthermore, this paper summarizes the research progress of semantic text similarity measurement and develops a more comprehensive classification description system for text similarity measurement algorithms. The classification includes string-based, corpus-based, knowledge-based, deep learning-based, traditional pretraining-based, and state-of-the-art pretraining-based methods. For each method, this paper introduces typical models and methods and discusses the advantages and disadvantages of these approaches. The systematic research on text similarity measurement methods enables a quick grasp of these methods, summarizing and analyzing classic and the latest research in text similarity measurement. The paper also lists evaluation indicators in this field and concludes by discussing potential future research directions. The aim is to provide a reference for related research and applications.
Keywords:	NLP, Text Similarity, Semantic Similarity, Similarity Measurement, Deep Learning, Pretraining
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	HYBRID FEATURE-DRIVEN ENSEMBLE LEARNING IN ARABIC NLP: FUSING SEQUENTIAL NEURAL NETWORKS WITH ADVANCED TEXT ANALYSIS TECHNIQUES
Author:	EMAN ALJOHANI
Abstract:	This paper investigates the application of deep learning methodologies for Arabic news classification, with a primary focus on the role of text preprocessing in enhancing model performance. By evaluating stemming across a range of datasets, the study hopes to clarify how effective stemming is at enhancing classification results. This study provides an extensive comparative analysis of deep learning models' effects on Arabic text processing. We introduce a novel hybrid neural network that combines TF-IDF weighting and FastText embeddings with Sequential Networks layers for NLP text classification. This architecture uses both static and dynamic language features for improved classification, capturing both the temporal word dependencies and the granular semantics. This paper proposed the new Hybrid-Bi model, a sophisticated approach that combines hybrid feature, BiLSTM, BiGRU, SVM, XGBoost, and Random Forest using stacking. It consistently performs better than other approaches on a range of news sources with and without stemming. The data also reveals a tendency where models generally outperform stemmed models, suggesting that stemming may omit important semantic information necessary for precise interpretation and classification in Arabic NLP. It achieves peak accuracies of up to 0.98 in Arabiya and Khaleej, especially in no stemming scenarios.
Keywords:	Text Classification, NLP, Feature Extraction, Deep Learning, Ensemble Learning
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	HEXAGONAL IMAGE COMPRESSION BASED ON WAVELET-GLOBAL THRESHOLDING
Author:	I M. O. WIDYANTARA, D. A. K. PRAMITA, M. D. P. ASANA, N. PRAMAITA, I G. A. K. D. D. HARTAWAN, J. K. MANI
Abstract:	This paper proposes a hexagonal grid image compression framework based on Wavelet thresholding. The target is to obtain optimal Rate-Distortion (RD) performance by implementing a threshold value prediction algorithm based on Global thresholding method. The compression scheme is built in two steps: resampling the hexagonal grid image and compressing it using the Wavelet-Global thresholding. The resampling process is carried out by alternating row and column suppression methods and an image interpolation method based on the Gabor Filter. In the Wavelet-Global thresholding, the coefficients of Gabor image are transformed into the Wavelet Coiflet, and the coefficients are thresholded globally using the Global thresholding. Furthermore, the quantization method and Arithmetic coding are applied to obtain a Hexagonal grid image compression scheme. The performance evaluation is performed on the Hard and Soft thresholding functions. Based on the threshold value generated by the Global thresholding, the Hard and Soft thresholding functions will limit the value of the Wavelet coefficients and impact the image's compression ratio and visual quality. The simulation results show that the best RD performance of the hexagonal grid image compression framework is obtained when applying a 1st order Coiflet filter with the Soft thresholding function
Keywords:	Hexagonal Grid Images, Gabor filter, Coiflet Wavelet filter, Global thresholding, Hard and Soft thresholding, Arithmetic coding, Rate-Distortion performance
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	BTSF: A BLOCKCHAIN TECHNOLOGY ENABLED SECURITY FRAMEWORK FOR STORAGE AND RETRIEVAL OF HEALTHCARE DATA BY INCORPORATING SMART CONTRACTS
Author:	G. JITHENDER REDDY, T. UMA DEVI
Abstract:	Blockchain technology enabled different real world applications to have secure storage and retrieval complying with desired security attributes. Of late, researchers focused on using blockchain for healthcare domain. Since health data is sensitive in nature and it needs technology-driven approach for data security and integrity. Many existing methods found in literature revealed their utility with blockchain integration. However, there is need for a comprehensive system with support for all operations controlled by smart contracts. Besides, the application should be user-friendly for user who does not know underlying technicalities. In this work, we suggested a framework named Blockchain Technology-enabled Security Framework (BTSF). Solidity language is used to define smart contracts. Two algorithms are proposed to realize the functionality of BTSF. They are known as Blockchain-enabled Security for Health Data Storage (BS-HDS) and Blockchain-enabled Security for Health Data Retrieval (BS-HDS). These two algorithms ensure secure healthcare data storage and retrieval. A prototype application is developed to evaluate our framework and underlying algorithms. Empirical study showed that BTSF is highly secure rendering non-repudiation and data integrity.
Keywords:	Security, Blockchain Technology, Smart Contracts, Healthcare Data Security, Secure Data Storage, Secure Data Retrieval
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	A COMPREHENSIVE EXAMINATION OF USER SATISFACTION IN INDONESIAN PASSPORT SERVICES: INSIGHTS FROM THE M-PASPOR APPLICATION
Author:	MAHESWARA RABBANI, AHMAD NURUL FAJAR
Abstract:	Measuring user satisfaction with the mobile application (M-Paspor) for making passports is a challenge currently being faced by public services from the Directorate General under The Ministry of Law and Human Rights of the Republic of Indonesia (Kemenhumkam). Fueled by a surge in demand for passports, the research aims to identify factors influencing user satisfaction and proposes improvements for a seamless user experience. The results of a survey of 100 users of the M-Paspor application indicate that system quality and perceived value have a positive and significant effect on user satisfaction, while information quality, service quality, and perceived ease of use do not have a significant effect. The research results show that system quality, information quality, and service quality are mediated by perceived value variables that influence user satisfaction. Other findings reveal that navigability and responsiveness, but not download delay, have a positive and significant effect on perceived ease of use. In conclusion, the study recommends ongoing enhancements to the application's features and user interface, incorporating user feedback for continuous optimization. The findings provide guidance and contribute valuable insights in measuring the success of the digital public service sector.
Keywords:	E-government, Indonesia, M-Passport Application, Public Service, User Satisfaction
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	A COMPARATIVE STUDY OF AI-BASED EDUCATIONAL TOOLS: EVALUATING USER INTERFACE EXPERIENCE AND EDUCATIONAL IMPACT
Author:	Dr. AMIRAH ALGAHTANI
Abstract:	This study provides a comprehensive analysis of AI-based educational tools, focusing on their impact on user experience and education. It explores the capabilities of AI tools in transforming the teaching and learning process through specialized AI tool-based learning, intelligent educational AI systems, Automation in the grading process, and predictive analytics. This research helps investigate the role of large language models (LLMs) in educational assessment, including test planning, item generation, test administration, and scoring. It involves teachers with STEM-related teaching experience who were introduced to an AI-enhanced scaffolding system for scientific writing. The study also includes a systematic review of AI applications in higher education, highlighting the ethical implications, challenges, and risks associated with AI in education. The findings provide a deep dive for educators, management, and stakeholders working on maximizing the outputs of AI in education while eliminating the associated risks. The study emphasizes the importance of understanding teachers' attitudes and experiences with AI in education to effectively integrate AI into teaching and learning practices. It also highlights the need to further explore ethical and educational approaches to applying AI in education. The research underscores the benefits and challenges of AI integration in education, emphasizing the need for transparent and ethical AI algorithms, personalized and adaptive assessment approaches, and the importance of human judgment in AI-powered education.
Keywords:	Educational Tools, AI In Education, Skills, Availability, Reliability
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	NOVEL SEGMENTATION BASED CERVICAL CANCER DETECTION USING DEEP CONVOLUTIONAL BASED NEURAL NETWORK WITH RELU
Author:	SOUMYA HARIDAS, DR. T. JAYAMALAR
Abstract:	Malignant growth in the cervical area is the fourth most common cause of death in women worldwide. The global burden of cervical cancer has decreased as a result of early screening, which made the disease a preventable one. Early detection and treatment of this cancer may reduce its adverse effects. In this paper, a proposed Deep Convolution-based neural network is used to find cervical cancer. The cervical image is preprocessed utilizing the Anisotropic Diffusion Filter (ADF), in which the edges of the image get preserved. Dragonfly optimization (DA) is used to optimize the weights of ADF. The weighted Fuzzy C-Means (WFCM) clustering method is utilized for segmentation, and makes the weight as optimized in WFCM by using the Grasshopper Optimization Algorithm (GOA). Consequently, a Deep Convolutional Neural Network (Deep CNN) employing a Rectified Linear Unit (ReLU) as the activation function is utilized for the extraction and classification of features. The Deep CNN surpasses alternative classifiers, achieving an accuracy of 97.8% in identifying cervical cancer, as evidenced by the study's findings on the performance of the proposed method relative to existing classifiers.
Keywords:	Cervical Cancer, Anisotropic Diffusion Filter (ADF), Dragonfly optimization (DA), weighted Fuzzy C-Means (WFCM), Grasshopper Optimization Algorithm (GOA), Deep Convolutional Neural Network (Deep CNN), Rectified Linear Unit (Relu).
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	COMBINATION OF CASE-BASED REASONING AND CERTAINTY FACTOR FOR DETECTION OF BABY BLUES SYNDROME
Author:	HANDRIZAL, DESILIA SELVIDA, SHAFIRA ALFINA
Abstract:	Baby blues is a mental health disorder that has recently become a public concern. Baby blues is a common emotional condition experienced by most young mothers in the days or weeks after giving birth to a baby. This condition is caused by hormonal changes that occur during pregnancy and after giving birth. To facilitate and reduce or prevent Baby Blues Syndrome, experts created an application program to provide solutions for mothers who experience Baby Blues Syndrome through an early detection process and providing appropriate treatment recommendations. By using the Certainty Factor and Case-Based Reasoning algorithms. Case-based reasoning will find the similarity value of new cases with existing cases, and the Certainty Factor to find the certainty value of the damage experienced by calculating the weight value. Both of the algorithms will result in a system that is more robust and can handle both uncertainty and utilize experience from previous cases. The research was carried out with 30 tests, with 13 questions, and resulted in a research accuracy value is 90%.
Keywords:	Baby Blues Syndrome, Expert System, Case-Based Reasoning, Certainty Factor
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	ANALYSIS OF PAPR, BER AND CHANNEL ESTIMATION IN MULTI CARRIER MODULATION SYSTEMS USING NEURAL NETWORKS
Author:	C RAMAKRISHNA , Dr S. SRINIVAS , DR.NAGARJUNA REDDY ADAPALA , AMDEWAR GODAVARI , SWAPNA SUNKARA , CHINTA NAGARAJU , KONDA MANOJ KUMAR , DR.YELIGETI RAJU , KARTHIK KUMAR VAIGANDLA , RADHAKRISHNA KARNE
Abstract:	The interference cancellation (IC) technique is a good choice for channel estimation (CE) in orthogonal frequency division multiplexing (OFDM) and Filter Bank Multicarrier (FBMC) systems due to the fact that it has a high level of accuracy in CE. FBMC is a crucial mechanism employed in 5G networks to optimize the available bandwidth while satisfying the demands for high spectral efficiency (SE). It is a feasible substitute for the OFDM modulation technique. The primary objective of this article is to examine the process of CE and IC, peak to average power ratio (PAPR) and bit error rate (BER) analysis in FBMC. Neural networks (NNs) are employed to approximate the optimal channel and retrieve the accurate transmitted signal with a minimal BER. We employ scattered pilots in both the time and frequency domains to estimate the channel for doubly-selective channels (DSC). Additionally, we utilize low-complexity IC techniques. The proposal for CE and IC includes the implementation of NN. The output sequences generated by the CE and IC algorithms serve as inputs for the NN. The results demonstrate that the proposed strategy closely approximates the ideal channel and exhibits a better BER performance compared to previous methods. This approach almost enhances the accuracy of CEs and significantly reduces the computational complexity (CC) in 5G networks.
Keywords:	BER, channel estimation(CE), Deep learning(DL), FBMC, interference cancellation(IC), MMSE, OFDM, PAPR, Recurrent neural network(RNN), LS, LSTM.
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	ENERGY EFFICIENT ROUTING USING SUPPORT VECTOR MACHINE IN WIRELESS SENSOR NETWORKS
Author:	GANTA CHAMUNDESWARI, B.VEERAMALLU, CHAGANTI B N LAKSHMI, RAVI AAVULA
Abstract:	Cluster head selection and energy utilization are efficiently managed using a conventional routing mechanism employing Wireless Sensor Nodes (WSNs). The main objective of the paper is to enhance network lifetime with average greater energy utilization. The Support Vector Machine (SVM) is used to tackle routing problems in the mobile base station connected with the infrastructure network. The protocol is intended to avoid the control by a centralized router or mobile base station of the complete mobile sensor nodes. In comparison to traditional energy efficient algorithms, the validation of SVM methodology shows an effective routing efficacy. The results against typical routing techniques over WSNs have been found successful.
Keywords:	Energy Efficiency, SVM, Machine learning, Routing, Wireless Sensor
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	FN-AN: FREQUENT NODE BEHAVIOR ANALYSIS USING AUDITOR NODE FOR INTRUSION PREVENTION IN NETWORK CODING ENABLED SMALL CELLS
Author:	CHANUMOLU. KIRAN KUMAR, NANDHA KUMAR
Abstract:	New vulnerabilities exist in wireless networks because of their exceptional design, which are deficient in traditional wired networks. As a result, advanced Intrusion Detection and Prevention Systems have become a necessity in today's modern information infrastructure. Mobile small cell technology is seen as a 5G enabling technology because of its potential to efficiently and cheaply bring ubiquitous 5G services to users. In addition, Network Coding (NC) technology can be expected to be an advantageous option for the wireless infrastructure of mobile small cells to boost its throughput and functionality. However, mobile small cells that use NC are susceptible to pollution attacks and Denial of Service (DoS) attacks due to the shortcomings of NC itself. The flexibility and portability of mobile small cells offered by NC is seen as a viable technology for 5G networks that may encompass the metropolitan environment on demand. Despite the many advantages that NC-enabled mobile small cells provide to the 5G of mobile networks, they present substantial security threats, which take advantage of NC's inherent vulnerabilities. Therefore, in order for NC-enabled mobile small cells to function to their full potential, intrusion prevention methods to identify and neutralize attacks are of the utmost necessity. In common parlance, an intrusion is any form of unauthorized intervention, which is almost often done maliciously. The goal of an intrusion is to gain access to an organization's internal network so that malicious actors can gather intelligent information about the organization, such as the layout of its networks or the types of software it uses, such as the operating system, tools, or applications. The Intrusion Prevention System (IPS) is sometimes known as a Intrusion Detection System (IDS), or Intrusion Detection/Prevention System. It is a program designed to keep NC enabled small cells safe by monitoring the suspicious behavior of nodes in the network. Intrusion prevention systems' primary roles include detection, analysis, reporting, and prevention of harmful behavior. Blockchains are decentralized databases that consist of continuously expanding lists of entries called blocks that are cryptographically linked together. Each block includes transaction data, a timestamp, and a cryptographic hash of the prior block. This research can make use of blockchain for recording the transactions occurred in the network. The node behavior can be stored in a block that is used for detection of malicious nodes easily in further transactions. This research presents a Time Frequent Node Behavior Analysis using Auditor Node with Flag Variable based Intrusion Prevention System (TFNBA-ANFV-IPS) with block register module using blockchain for accurate detection and prevention of intrusions. The proposed model when contrasted with traditional models performs better performance in intrusion prevention.
Keywords:	Network Coding, Small Cells, Intrusion Detection, Intrusion Prevention, Attack Detection, Node Behavior, Blockchain, Auditor Node.
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	A MULTIOBJECTIVE EVOLUTIONARY ALGORITHM WITH A NOVEL MUTATION OPERATOR FOR OVERLAPPING COMMUNITY DETECTION
Author:	A.C. RAMESH, G. SRIVATSUN
Abstract:	The detection of overlapping communities in complex real-world networks is a difficult problem that is being addressed by different methods. The multiobjective evolutionary algorithm (MOEA) is a promising alternative that has shown competitive performance in this research over the past two decades. The representation scheme used by the MOEA affects the quality of the solutions obtained and the runtime of the algorithm. The length of the chromosome is significantly reduced when cliques are used as genes instead of the original nodes of the graph. The execution time of the evolutionary algorithm (EA) also depends on the combined execution times of the evolutionary operators, crossover and mutation. This paper proposes a novel mutation operator that uses community labels of cliques as genes rather than cliques. The proposed mutation operator results in fewer modifications on the chromosome than does the existing clique-based mutation. Experiments conducted on real-world and synthetic networks reveal that the proposed algorithm produces good community partitions of the network when compared with existing clique-based algorithms and state-of-the-art community detection algorithms.
Keywords:	Community-Based Mutation, Maximal Clique, Quantile, Overlapping Community Detection, Multiobjective Evolutionary Algorithm.
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	MACHINE LEARNING APPROACHES FOR HELPDESK TICKETING SYSTEM: A SYSTEMATIC LITERATURE REVIEW
Author:	ALVIAN SHANARDI WIJAYA, TANTY OKTAVIA
Abstract:	Machine learning has been commonly used as a tool to support helpdesk function in many areas especially in ticketing system area. Ticketing system is one of the most common features for Helpdesk to provide support for users from answering question, resolving error and giving feedback to the products or services provided by the company. Combined with machine learning, helpdesk could classify user’s problem according to their impact and urgency level by learning datasets produced helpdesk. This study aims to review relevant works about machine learning approaches in various use case scenario in helpdesk ticketing function. This study will perform a systematic literature review using PSALSAR Framework as a tool to study this knowledge based on SCOPUS database starting from 2012 until 2022. This paper will specifically search for three keywords (“Machine Learning”, “Helpdesk” and “Ticketing”) to find the related article and present a systematic literature review using three research question. Those findings will be discussed using quantitative, descriptive and narrative analysis to answer all of research question by manually assessed and extracting necessary data from each individual study This paper will produce systematical review of machine learning use cases and method in helpdesk ticketing function which could be useful for further research in other helpdesk ticketing function that has not been researched and provide analytical data about machine learning limitations and what could be done in the future to create more advanced machine learning model in helpdesk ticketing systems.
Keywords:	Machine Learning, Helpdesk, Ticketing System, Literature Review
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	STREAM OF TRAFFIC SIGN RECOGNITION BY MEANS OF AUTO ENCODER CLASSIFICATION MODELS
Author:	DR. PANCHAGNULA VENU MADHAV, DR. SURESH KUMAR PITTALA, ASHOK KUMAR KAVURU, A MURALI KRISHNA, BALAJI TATA, KURRA UPENDRA CHOWDARY, N.JAYA
Abstract:	Raising performance standards by carefully combining tried-and-true techniques with cutting-edge approaches. Based on the foundation of the current YOLOv5 algorithm, which is well-known for its object identification skills, this paper aims to improve its performance by combining it with new models, such as the Autoencoder- frameworks. Through combining these disparate methods, the study seeks to use each of their unique advantages, ultimately resulting in a thorough comparison study that reveals their separate effects on precision and productivity. In addition to improving traffic sign recognition systems' accuracy, this methodical assessment—which is characterized by rigorous testing and strong optimization—also reveals illuminating connections between the suggested and established methods.. The main goal of this endeavor is to unravel how these seemingly unrelated components, when brought together, can potentially usher in a new age of higher performance standards. This work aims to create a route towards the development of more sophisticated, flexible, and well-tuned traffic sign detection and identification systems by bridging the gap between the established and the cutting edge, with consequences covering a variety of real-world applications.
Keywords:	Stream of Traffic Sign, Frameworks, Autoencoder, Comprehensive
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	MODELING THE EFFECTS OF SELF-LEARNING AND KNOWLEDGE SHARING IN OER COURSES AMONG COLLEGE STUDENTS
Author:	ABDULLAH AHMED ELNAQLAH, MERZA ABBAS
Abstract:	The purpose of the study was to investigate whether Boisot’s (1998) Social Learning Cycle was applicable in describing the effects of Self-Learning and Knowledge sharing on Transformation of the learner in an OER environment. Also investigated were whether there were significant differences by demographic factors for gender and year of study for Self-learning, Knowledge sharing and Transformation of the learner. This study employed the survey research method and a questionnaire comprising 37 Likert-scale items involving seven factors was adapted for this study and administered online. 144 respondents from colleges that actively employed OER courses in Palestine completed the survey. Data was analyzed using SmartPLS to verify the proposed model and ANOVA to compare learning engagement scores by the demographic factors. The findings showed that the hypothesized model derived from Boisot’s Social Learning Cycle model fitted the data with Self-Learning having significant direct effects on Transformation at β = 0.359 and on Knowledge Sharing at β = 0.665. Also, Knowledge Sharing reported a significant direct effect on Transformation with β = 0.472, giving an indirect effect of Self-Learning on Transformation at β = 0.314. In addition the findings showed that Knowledge Sharing reported a partial mediation effect of 46.60% and the effect size of Self-Learning on Transformation was at f 2 = .159 or of medium effect size while the effect size for predictive relevance was large at q2 = .378. The ANOVA findings reported that there was a significant difference in knowledge sharing by Year of Study with 4th year students reporting significantly higher scores than 1st year students, but there were no significant differences for other factors of the model and by Gender. These findings indicated that students in Palestine were actively involved in knowledge sharing when engaging in OER courses and the students benefitted more when they participated in knowledge sharing than by studying alone.
Keywords:	Self-Learning, Knowledge sharing, OER, Social Learning Cycle, Palestine
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	OPTIMIZATION OF SMART TAXATION USING ARTIFICIAL INTELLIGENCE : RISKS AND OPPORTUNITIES
Author:	YOUSRA ABKAR , SAIDA CHAIHAB
Abstract:	The article explores the impact of artificial intelligence (AI) on effectiveness of management control, highlighting its transformative role in finance. The introduction highlights the growing importance of AI in finance, highlighting its benefits in data collection, AI model development, and performance evaluation. The article's development examines data collection in detail, highlighting the use of historical and real-time sources to power AI models. Model development methods are presented, highlighting approaches such as neural networks, SVMs, genetic algorithms, natural language processing, reinforcement learning models, decision tree ensembles, and semantic knowledge networks. AI in finance transcends the boundaries of technological innovation, optimizing decision-making, strengthening financial security, improving operational efficiency, and personalizing financial services for customers. Developmental paragraphs highlight the importance of AI in fraud detection and prevention, process automation, and personalization of financial services. A detailed section explores the different AI models applied to portfolio management, such as neural networks, SVMs, genetic algorithms, natural language processing-based models, reinforcement learning models, sets of decision trees, and semantic knowledge networks. Comparing the results obtained by AI models with traditional approaches highlights the transformation of the financial landscape towards more sophisticated and adaptive methods. We present the emerging context of smart taxation and highlight the importance of research as an essential driver of its development. By highlighting current gaps in traditional approaches to taxation, we introduce the need to explore new AI-based methods to improve the efficiency of the system. The article then analyzes the advantages and limitations of the proposed models, highlighting the ability of AI models to process massive volumes of data in real time and dynamically adapt to market changes. The conclusion summarizes the contributions of AI to portfolio management, highlighting its major role in optimizing financial strategies.
Keywords:	Artificial Intelligence, Finance, Portfolio Management, Financial AI Models, Strategic Optimization, AI Model Performance, SVM
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	ADVANCES IN MACHINE LEARNING AND DEEP LEARNING APPROACHES FOR PLASTIC LITTER DETECTION IN MARINE ENVIRONMENTS
Author:	ABDELAADIM KHRISS, AISSA KERKOUR ELMIAD, MOHAMMED BADAOUI, ALAE-EDDINE BARKAOUI, YASSINE ZARHLOULE
Abstract:	A serious threat to the environment is plastic pollution in marine ecosystems, and thus an effective detection of litter plastics is needed for proper management. This review critically assesses recent studies that use CNNs and other machine learning approaches to detect and measure plastic debris in various water bodies. The study delves into the models, datasets, and evaluation measures used in these studies factoring in persistent challenges associated with detecting small objects and variability of environmental conditions. In addition, the study offers future perspectives highlighting the need for complete data gathering, utilization of various sources of imagery, and development of real-time monitoring mechanisms to combat plastic pollution. Through the integration of these findings, this review attempts to assist researchers, decision-makers, and stakeholders in designing creative approaches for minimizing the destructive consequences of plastic pollution on marine environments.
Keywords:	Machine Learning, Deep Learning, Object Detection, Remote Sensing, Monitoring
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	BIDIRECTIONAL CNN-LSTM ARCHITECTURE TO PREDICT CNXIT STOCK PRICES
Author:	PRIYANKA DASH, JYOTIRMAYA MISHRA, SURESH DARA
Abstract:	Stock price prediction has long been a central concern for investors and financial analysts. This research paper explores applying a Bidirectional Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) architecture to predict stock prices, specifically focusing on the CNXIT (Nifty IT) stock index. The study investigates the potential of deep learning techniques to capture complex temporal dependencies and spatial patterns in historical stock price data. The research begins with a comprehensive review of existing literature on stock price prediction and the utilization of deep learning methodologies. It introduces the Bidirectional CNN-LSTM model and details the data preprocessing steps, model architecture, and training process. The dataset, comprising historical CNXIT stock prices, is meticulously cleaned and prepared to ensure the model's accuracy. Experimental results and findings demonstrate the model's predictive performance, including metrics such as mean squared error (MSE), mean absolute error (MAE), and explained variance. Visualizations of the model's predictions alongside actual CNXIT stock prices offer valuable insights into its ability to anticipate market trends. The paper concludes by discussing the implications of the Bidirectional CNN-LSTM architecture in stock price prediction and its potential to enhance decision-making in financial markets. Future research directions and areas for model improvement are also explored. This research contributes to the evolving landscape of financial forecasting by showcasing the efficacy of Bidirectional CNN-LSTM in predicting stock prices within the context of the CNXIT index.
Keywords:	CNN, LSTM, MSE, CNXIT, Prediction, Deep learning
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	HARNESSING REINFORCEMENT LEARNING IN FOG-CLOUD COMPUTING: CHALLENGES, INSIGHTS, AND FUTURE DIRECTIONS
Author:	MUSTAFA AL-HASHIMI, AMIR RIZAAN RAHIMAN, ABDULLAH MUHAMMED, NOR ASILAH WATI HAMID
Abstract:	The fast-changing world of fog-cloud computing poses various challenges and opportunities, especially in terms of optimizing resources, adaptability, and system efficiency. Reinforcement Learning (RL) is a powerful tool to tackle these challenges due to its ability to learn and adjust from interactions. This article explores the different RL algorithms, emphasizing their distinct strengths, weaknesses, and practical implications in fog-cloud environments. We present a comprehensive comparative analysis, from the deterministic nature of Q-Learning to the scalability of DQN and the adaptability of PPO, providing insights that can assist both practitioners and researchers. Additionally, we discuss the ethical considerations, real-world applicability, and scalability challenges associated with deploying RL in fog-cloud systems. In conclusion, while integrating RL in fog-cloud computing shows promise, it requires a comprehensive, interdisciplinary approach to ensure that advancements are ethical, efficient, and beneficial for everyone.
Keywords:	Fog-Cloud Computing, Q-Learning, Deep Deterministic Policy Gradient (DDPG), Proximal Policy Optimization (PPO), Deep Q-Network (DQN), Reinforcement Learning.
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	UNVEILING ANOMALIES IN CROWDS THROUGH ADVANCED DEEP LEARNING FOR UNUSUAL ACTIVITY DETECTION
Author:	MUTHURASU N, RAJASEKAR V
Abstract:	Abnormal activity in the modern environment suggests risks and threats to other people. Anomaly refers to anything that differs from what is typical, anticipated, or normal. Given the challenges of consistently monitoring public areas, the implementation of intelligent video surveillance is imperative. Detecting unusual crowd activities is a complex subject that has spurred research advancements in the field of surveillance video applications. The main objective of this research is to identify atypical gatherings, instances of anomalous crowd behavior. Various techniques, including histogram representation, optical flow calculation, and deep learning-based algorithms, have been employed to address these issues. Nevertheless, there is a deficiency in effectively addressing this issue due to blockage, noise, and congestion. The introduction of AI techniques resulted in significant technological advancements. During the real-time monitoring of video material, the system employs various techniques to differentiate between different suspicious activities. The unpredictability of human behavior makes it challenging to discern whether it is suspicious or typical. Conducting monitoring commonly involves pulling consecutive frames from a video. There are two components in the framework. During the initial stage, the framework computes the features from the video frames. In the subsequent step, the classifier utilizes these features to determine if the class is panic or normal. The suggested methodology is evaluated using three available datasets, namely PETS 2009, MED and UMN dataset. The suggested method is compared with existing techniques to assess its efficiency.Abnormal Activity, Anomalous, Video Surveillance, Deep Learning, Crowd Behavior, Video Frames, PETS, UMN Dataset.
Keywords:	Abnormal Activity, Anomalous, Video Surveillance, Deep Learning, Crowd Behavior, Video Frames, PETS, UMN Dataset.
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	HEART ATTACK PREDICTION USING MACHINE LEARNING: A COMPREHENSIVE SYSTEMATIC REVIEW AND BIBLIOMETRIC ANALYSIS
Author:	JAVIER GAMBOA-CRUZADO, RENZO CRISOSTOMO-CASTRO, JHONATAN VILA-BULEJE, JEFFERSON LÓPEZ-GOYCOCHEA, JORGE NOLASCO VALENZUELA
Abstract:	Studies on predicting heart attacks using Machine Learning demonstrate that there is a wide variety of algorithms and methodologies highlighting their impact on heart attack prediction. This can help in reducing the risk of lifestyle-related complications. To understand the current state of the art, a systematic literature review (SLR) was conducted from 2017 to 2021. A key step in this SLR was the search strategy, which identified 3,525 articles from various sources of information such as Taylor and Francis, IEEE Xplore, ARDI, ACM Digital Library, ProQuest, Wiley Online Library, and Microsoft Academic. Exclusion criteria were applied, such as articles older than five years, non-English articles, and papers not published in conferences or journals, to ensure only the most relevant studies were included, ultimately resulting in 82 articles. The findings from the systematic review focused predominantly on studies predicting heart attacks, detailing the best methodologies and algorithms used to enhance the accuracy of these predictions. The conclusions indicate that, despite different approaches, the articles exhibit common themes and objectives in achieving better heart attack predictions using Machine Learning.
Keywords:	Heart Attack Prediction, machine learning, cardiac problems, ML, cardiac disease, Systematic Literature Review
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	DIABETIC RETINOPATHY CLASSIFICATION WITH DEEP NETWORKS AND FEW-SHOT LEARNING ON IMBALANCED RETINAL DATA
Author:	N V NAIK, HYMA J, PVGD PRASAD REDDY
Abstract:	A consequence of diabetes called diabetic retinopathy (DR) has been caused by damage to the vessels of blood that carry photosensitive cells in the eyes. If this issue is not identified in its early stages, vision loss follows. Diabetic retinopathy occurs in five stages: proliferative, mild, moderate, severe, and no DR. Conventional approaches for DR detection are time-consuming. The availability of sufficient data for training would be necessary for an autonomous and accurate model, however, this is not the case. Aside from No DR, the publicly accessible dataset is wildly unbalanced for other classes, particularly proliferative and severe classifications. The performance of Deep Learning (DL) models has been negatively impacted by issues with imbalanced datasets, inconsistent annotations, fewer sample images, and improper performance evaluation measures. The current state-modern-the-art DL (deep learning) techniques, particularly CNN architectures, have been applied to numerous issues and have demonstrated remarkable efficacy in learning balanced datasets. Furthermore, a sizable dataset is needed to train the model using DL-based approaches. The main obstacle to creating deep learning models is the sheer volume of data that is unavailable, particularly for uncommon and emerging retinal disorders. few-shot When a sizable dataset is not available for training, learning can be used as a substitute for developing deep learning models. To solve the aforementioned problems, the researchers in this study established DR-FSL-DNet, a FSL-based deep network model for DR classification. Few-Shot learning (FSL) produced better results than DL models by guiding the DR classification model with a comparatively limited number of samples. First, The DR-FSL-DNet framework utilizes episodic learning to train its model on few-shot classification tasks. Later, applied resampling technique (i.e., Focus Loss function) to balance data and classes in training data. Finally, a prototype meta-learning network is applied for DR detection and classification. On public datasets DRIVE, STARE, and CHASEDB1, the proposed network DR-FSL-DNet is applied by comparing it with modern-of-the-art works. The experiment shows that DR-FSL-DNet contributes to the desirable performance of DR classification. The proposed network outperforms compared to standard methods and achieved better scores in Accuracy, Sensitivity, Specificity, Precision, and F1-Score metrics.
Keywords:	Diabetes Retinopathy, Deep Learning, FSL, Class Imbalance, and Retinal Data.
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	AI DRIVEN GAME THEORY OPTIMIZED GENERATIVE CNN-LSTMMETHOD FOR FAKE CURRENCY DETECTION
Author:	MS.K. SWEETY, DR. M.NAGALAKSHMI, MS.RAHAMA SALMAN, DR GANTA JACOB VICTOR, ASLAM ABDULLAH M, PROF. TS.DR.YOUSEF A.BAKER EL-EBIARY
Abstract:	Imitation money, or fake cash, is a serious danger to global financial systems' reliability and security. Such currency items are made illegally with the intention of misleading people, organizations, and governments. Identifying counterfeit money is essential for a number of reasons, chief among them being the preservation of trust between the currency as well as banking systems. Fake money may upset economic stability, cause financial losses for people and enterprises, and destroy trust within the monetary system. Consequently, there may be a decrease of GDP and a rise in living costs. The effective operation of a nation's economy depends on the detection of counterfeit currency as it protects the integrity of a country's currency, prevents fraud, and preserves the safety of money transactions. The most advanced strategy for preventing counterfeit money is the AI-driven Game Theory Optimized Generative CNN-LSTM technique for Fake Currency Identification. The ongoing issue of counterfeiting calls for sophisticated and flexible solutions. This approach integrates the use of Generative Adversarial Networks (GANs) using game theory optimization, Long Short-Term Memory (LSTM) networks for temporal pattern recognition, as well as Convolutional Neural Networks (CNNs) for feature extraction. The device that discriminates has to attempt to separate synthetic pictures of counterfeit cash produced by the GAN from real banknotes. The technique is extremely accurate and flexible when it comes to identifying fake money. It offers greater possibilities that might improve safety within banking environments and other areas. The proposed framework is implemented in python. The proposed Game Theory Optimized Generative CNN-LSTMMethod shows better accuracy with 98.9% when compared with SVM, AlexNet, and Linear Discriminant Analysis.
Keywords:	Convolutional Neural Network (CNN); Fake Currency; Generative Adversarial Network (GAN); Game Theory Optimization;Long Short-Term Memory (LSTM).
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	INTEGRATION OF IOT-ENABLED TECHNOLOGIES AND ARTIFICIAL INTELLIGENCE IN DIVERSE DOMAINS: RECENT ADVANCEMENTS AND FUTURE TRENDS
Author:	NISSRINE GOUIZA, HAKIM JEBARI, KAMAL REKLAOUI
Abstract:	The concept of the Internet of Things (IoT) has risen as a revolutionary innovation, establishing a connection between the physical and digital worlds and significantly impacting various aspects of daily life. In the healthcare field, it has unlocked the potential of connected medical devices, enhancing care through real-time patient monitoring and the effective management of chronic diseases. Within the industry, IoT facilitates predictive maintenance, optimizes manufacturing processes, oversees the supply chain, and monitors assets. Smart cities utilize IoT to elevate infrastructure management, enhance security, and promote sustainability. In agriculture, IoT sensors bring about a transformation in precision farming, optimizing resource utilization, and increasing yields. Smart homes integrate IoT for home automation solutions, empowering homeowners to remotely control devices and systems. Finally, in the transportation, IoT is at the forefront of revolutionizing connected and autonomous vehicles, providing advanced features in safety, navigation, and onboard entertainment. The integration of (IoT) and (AI) yields considerable benefits across various sectors by enhancing operational efficiency, facilitating informed decision-making, and fostering the creation of smarter, interconnected environments. In this article, we conducted a bibliometric study focused on industrial sectors related to the Internet of Things (IoT) from 2018 to 2023. Our analysis centers on comparing the most frequently explored domains, highlighting their popularity and performance. Furthermore, we examined currently predominant and beneficial technologies, particularly those aimed at optimizing operations, improving efficiency, and reducing costs.
Keywords:	IOT, AI, AIOT, IIOT, Agriculture, Aquaculture, Healthcare, Transportation, Smart Homes, Smart Cities, Real-Time Monitoring, Sustainability.
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	AN ANONYMOUS MUTUAL AUTHENTICATION MECHANISM FOR WEARABLE SENSORS IN THREE-TIER MOBILE HEALTHCARE SYSTEMS
Author:	A. HEMLATHADHEVI, D. R. THIRUPURASUNDARI, C. RAMESH KUMAR, G. NIRMALAR
Abstract:	In light of the openness and mobility of wireless communication, Mobile Healthcare Systems are vulnerable to a wide range of threats, which considerably reduces their value and hinders their widespread implementation. Patients and medical personnel can be linked to their actions by attackers and criminals, even if they don't realize the context of the data they're transmitting, by simple eavesdropping. All levels of the mHealth ecosystem are affected by these flaws. This research proposes an anonymous mutual authentication mechanism for wearable sensors in three-tier mobile healthcare systems. The HSP medical server, controller nodes, and the anonymous authentication nodes of mobile users are all supported. It also makes it possible for mobile users and controller nodes to exchange authentication information anonymously. The controller nodes and the wearable body sensors are anonymously authenticated with the help of this method of authentication. In order to ensure the security of our protocols, we do extensive formal demonstrations and informal conversations about security features, prospective attacks, and responses. Simulated outcomes indicate that our strategy is safe and meets all of the required privacy and authentication requirements.
Keywords:	Wearable Sensors, Anonymity, Authentication, Healthcare, Security, Wireless Body Sensor Networks.
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	A FUSED 3D-2D CONVOLUTION NEURAL NETWORK FOR SPATIAL-SPECTRAL FEATURE LEARNING AND HYPERSPECTRAL IMAGE CLASSIFICATION
Author:	MURALI KANTHI, J. DAVID SUKEERTHI KUMAR, K. VENKATESHWARA RAO, MOHMAD AHMED ALI, SUDHA PAVANI K, NUTHANAKANTI BHASKAR, T. HITENDRA SARMA
Abstract:	Hyperspectral image (HSI) classification is a prominent topic in the area of remote sensing and it is challenging task since the minimal number of labelled training samples and the high dimensional space that includes a wide range of spectral bands. Therefore, a more effective neural network architecture needs to be created to boost the effectiveness of the HSI classification job. To tackle these issues, we present an innovative fused 3D-2D convolutional neural network (F-CNN), that extracts both spectral-spatial features and enhances the HSI classification efficiency by incorporating three fusion blocks sequentially into the proposed model. Each fusion block includes a module of 3D-2D CNNs to retrieve and fuse the spectral-spatial information for the improvement of HSI classification task. Three standard datasets that are freely available (Salinas, Pavia University, and Indian Pines) and new Indian datasets (Ahmedabad-1 and Ahmedabad-2) are carried out in experimental investigation to determine the efficiency of the presented model. The proposed F-CNN model has achieved accuracies of 80.79% on the AH1 (Ahmedabad-1) dataset, 87.98% on the AH2 (Ahmedabad-2) dataset, 99.99% on the PU (Pavia University), 99.99% on the SA (Salinas), and 99.92% on the IP (Indian Pines) dataset. According to research findings, the provided model outperforms the remaining current techniques in terms of classification performance.
Keywords:	Hyperspectral Image, Deep Learning, Convolutional Neural Network, Classification, Spectral-Spatial Information
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	CHRONIC HEPATITIS-C DETECTION INTERPRETATION USING MACHINE LEARNING ALGORITHM AND WITH CLINICAL FINDINGS
Author:	K. VIDYA SAGAR., CH.SUMA , SURYA PRASADA RAO BORRA, M.SUMALATHA, JAMMALAMADUGU RAVINDRANADH, LAKSHMI RAMANI BURRA
Abstract:	Hepatitis - C infection is spreading very rapidly. The infection rate is alarming. The infected liver leads to liver fibrosis and then leads to cirrhosis. The fibrosis and cirrhosis stage are associate the human life. This paper considered CT images for better interpretation and classification. The RBG image is converted to grey scale image. The contrast levels of the image are improved using CLAHE algorithm. The image quality parameters AMBE, PSNR, MSE, MD and MAE are analysed. The contrast adjusted image is processed for banalization using thresholding mechanism then CNN classifier is applied to classify the severity of liver fibrosis, and cirrhosis. The liver profile clinical findings aspartate aminotransferase enzyme (AST) and alanine transaminase (ALT), INR and Albumin total bilirubin are correlated with the machine interpreted results. caudate right lobe ratio of CLD/ RLD is also estimated to interpret the severity of the fibrosis and cirrhosis. The proposed methodology is more reliable for meticulous interpretation of liver inflammation with hepatitis-C virus.
Keywords:	Hep-C Infection, CT Images, CNN Classifier, Liver Fibrosis, Liver Cirrhosis.
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	MODEL APPLICATION BASED ON FUZY LOGIC TSUKAMOTO WITH CERTAINTY FACTOR FOR EARLY DIAGNOSIS ON CORONA VIRUS (COVID-19)
Author:	DHAIFINA AULIA, AMALIA ZAHRA
Abstract:	COVID-19 is a new type of infectious disease caused by the coronavirus that first appeared due to an outbreak in Wuhan, China in December 2019. The process of controlling and preventing COVID-19 by calibrating early diagnosis in Non-Suspect Cases, Suspect Cases, and Close Contacts is still in doubt. In addition, COVID-19 also has many symptoms that make the diagnosis process by direct diagnosis by medical personnel and without involving technology often quite difficult. So, a model application is needed that can help and facilitate doctors and medical personnel in making an initial diagnosis of COVID-19. So, to overcome uncertainty problems such as lack of information, inaccuracy, doubt, and incomplete truth, this study applies a model by implementing Fuzzy Tsukamoto with Certainty Factor. In this study using variables such as traveling, body temperature, close contact, symptoms and diseases. Where each of these variables has a fuzzy set, namely very sure, sure, fairly sure, a little sure, don't know, and no for all symptoms, except for body temperature symptoms, which have cold, normal, and hot fuzzy sets. In addition, there are 17 rules that are formed for the diagnosis. The evaluation results using the Fuzzy Logic Tsukamoto method with Certainty Factor on the real diagnosis of health workers on 30 data samples can produce an accuracy of 80%. As such, this study has the potential to make a positive contribution to supporting the early diagnosis of COVID-19 patients, in the hope of reducing hesitation and enabling faster decision-making in the management of such cases.
Keywords:	Corona Virus (COVID-19), Early Diagnosis, Fuzzy Logic Tsukamoto
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	A DECISION TREE WITH FILTER ATTRIBUTE SELECTION MODEL FOR DESIGNING AN AUTOMATED JOB RECOMMENDER SYSTEM FOR CANDIDATE MATCHING THROUGH ONLINE RECRUITMENT
Author:	S. KRISHNA KUMAR, DR. E. RAMARAJ, DR. S. SANTHOSHKUMAR, DR. P. GEETHA
Abstract:	Due to globalization and Advancement in the Technology, the job vendors across the globe search for candidates having some specific talents to fulfill the job vacancy of their companies. The online recruitment system acts as a bridge for connecting the job seekers and job vendors through Internet across the globe. Most of the companies nowadays rely on online job recruitment websites to hire employees for their companies. Due to the lack of proper resume format and difficult in extracting the job-related attributes from the uploaded resume most of the job recruitment websites failed to match the suitable jobs for the registered candidates. This research proposed a Decision Tree with Filter Attribute Selection (DT-FAS) Model for designing a recommender system for online job recruiting environment. Most of the online recruitment websites have complex registration process and it failed to collect the relevant job description and job skills attributes from the employees and recruiters. This research designed a registration form in Google Forms to simplify the registration procedure. The link of the form is send to users through WhatsApp and Emails and the job seekers are allowed to enter their details and the data is collected as Google Spreadsheet. Feature Engineering is done on the collected dataset to select the relevant job matching attributes with the help of filter feature selection techniques such as Info Gain, Chi Square and Tree Based methods. Decision Tree Induction algorithm was implemented on the job seekers data having selected attributes. The job seekers were classified according to the association rules generated from the decision tree and the appropriate job matching their profile was identified. The results analyzed with the performance metrics showed that the decision tree built with the selected attributes of Information Gain method gives best result and correctly identified job matching attributes present in the database.
Keywords:	Online Job Recommender System, Filter Feature Selection Techniques, Decision Tree Induction Algorithm.
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	OVERCOMING CONFLICT RESOLUTION WITH ANDROID APPLICATION-BASED LEARNING: DESIGN, DEVELOPMENT, AND IMPLEMENTATION OF A CASE STUDY
Author:	LEONARDUS SANDY ADE PUTRA, F. TRIAS PONTIA WIGYARINTO, NEILCY TJAHJAMOORNIARSIH, EKA KUSUMAWARDHANI, VINCENTIUS ABDI GUNAWAN, AGUS SEHATMAN SARAGIH, LILIS SURYANI
Abstract:	In today's era technology greatly affects our lives both in terms of education and daily needs. Technology in the telecommunications sector is currently very influential on the sustainability of education, especially for school students in Indonesia. Students in Indonesia are currently faced with a conflict between individuals and groups related to religion, ethnicity, and culture. Problems that arise can cause fellow students to deviate and become a serious problem. Conflicts that occur among students are now increasingly worrying. This is due to conflicts between students that lead to physical violence. So there needs to be an education about conflict resolution that can help students understand a problem. This study specifically aims to design an e-learning application that contains conflict resolution material and religious, ethnic, and cultural diversity for students in Indonesia. So, the students will be able to get an education related to conflict and diversity to create students who have good morals and behavior. The application is designed using Android Studio software. The application has main features, including information features, reading collection features, multiple-choice evaluation features, case analysis evaluation features, and video features. The reading feature is equipped with a recording from the teacher explaining each slide to help students who have visual impairments understand the material more deeply.
Keywords:	E-Learning; Self-Learning Application; Android; Mobile Application; Conflict Resolution.
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	METAHEURISTIC ALGORITHM-BASED LOAD BALANCING IN CLOUD COMPUTING
Author:	B. EDWARD GERALD, DR.P. GEETHA
Abstract:	Cloud computing has emerged as a revolutionary paradigm for delivering computing resources and services on-demand. To ensure the efficient utilization of cloud resources and provide high availability and reliability to users, load balancing is a critical component. Load balancing aims to distribute incoming network traffic or computational tasks across multiple cloud servers, preventing overloading on specific servers and optimizing resource utilization. Traditional load balancing techniques, such as round-robin and least-connections, are often not sufficient to handle the dynamic and complex workload characteristics of cloud environments. In this context, metaheuristic algorithms have gained prominence as an effective approach to address the load balancing problem in cloud computing. This paper presents a comprehensive study on load balancing using metaheuristic algorithms in cloud computing. We explore the key challenges in load balancing for cloud environments and discuss how metaheuristic algorithms, including genetic algorithms, particle swarm optimization, and simulated annealing, have been applied to tackle these challenges. We investigate the theoretical underpinnings of these algorithms and their practical implications for cloud load balancing. Furthermore, we present a comparative analysis of the performance of various metaheuristic algorithms in different cloud computing scenarios. We evaluate their effectiveness in terms of reducing response time, optimizing resource utilization, and enhancing fault tolerance. Real-world experimental result is presented to illustrate the practicality and efficiency of metaheuristic-based load balancing solutions.
Keywords:	Load Balancing Techniques, Cloud Environment, Overloading, Metaheuristic Algorithm,
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	IOT-BASED SMART ENERGY MANAGEMENT SYSTEM TO MEET THE REQUIREMENTS OF EV CHARGING STATIONS
Author:	SWATHI DASI, S. MANI KUCHIBHATLA, MANAM RAVINDRA, KOMMERLA SIVA KUMAR, SUDHA SREE CHEKURI, ASHOK KUMAR KAVURU
Abstract:	Switchable building glazing powered by solar energy and Vanadium Redox Flow Battery (VRFB) technology is the subject of this research paper's investigation into an IoT-based smart energy management system. The main goal is to promote efficiency, maximize energy use, and assist the HVAC system of electric vehicle (EV) charging stations. Switchable building glazing, virtual reality field beam forming (VRFB) technologies, solar energy harvesting, and internet of things (IoT) smart energy management systems are all included in this paper's extensive literature analysis. Solar panel selection, VRFB setup, Internet of Things (IoT) sensor deployment, and integration with switchable building glazing are all covered in the methodology section, which also details the design, implementation, and testing process. Adaptability to changing energy demands in EV charging stations is emphasized in the discussion of the design and components of the IoT-based smart energy management system. This study investigates the feasibility of using VRFBs to collect solar energy surpluses for usage in HVAC and electric vehicle charging infrastructure. As an added bonus, we take a look at how switchable building glass may improve energy efficiency, thermal insulation, and natural light penetration. Challenges, improvement possibilities, and possible applications in different situations are addressed in the presentation and discussion of results from simulations and real-world testing. Highlighting the importance of the suggested approach for long-term, cost-effective solutions to electric vehicle charging infrastructure, the conclusion presents important results. Also included are suggestions for both academic and business use in the future.
Keywords:	Internet Of Things (Iot), Smart Energy Management System, Vanadium Redox Flow Battery (VRFB), Heating, Ventilation, And Air Conditioning (HVAC) system, electric vehicle (EV).
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	ADVANCEMENTS IN DEDUPLICATION TECHNIQUES FOR EFFICIENT DATA STORAGE
Author:	RICHA ARORA, D. VETRITHANGAM
Abstract:	Information deduplication is an arising innovation that presents decrease of capacity use and a Proficient method of taking care of information replication in Auxiliary Capacity. In the deduplication, information is separated into "different pieces" and exceptional hash identifier is related to each piece. These identifiers are used to contrast the pieces and recently put away lumps and checked for duplication. High throughput hash less lumping technique called Fast. The most extreme esteemed byte is remembered for the piece and situated at the limit of the lump. Huge measure of information gets produced consistently and putting away that information productively turns into a heuristic errand. Reinforcement stockpiles are more noticeably involved media for putting away consistently, the created information. The huge measure of information that is put away in the reinforcement stockpiling is excess and prompts the wastage of extra room. Extra room can be saved and handling velocity of reinforcement media can be further developed utilizing deduplication and variable size lumping. Different lumping calculations have been introduced in the past to further develop deduplication process. Foundation of information deduplication research includes a persistent work to address the developing difficulties in information capacity, investigate new advances, and streamline deduplication methods for different applications and conditions. This novel plan to find some kind of harmony between capacity productivity, execution, security, and versatility to meet the different requirements of present-day information the executives. This novel intent to provide the comparative study of various Data Deduplication techniques. It provides the best algorithm based on different backgrounds like Data Security, Performance etc. which can be further taken in consideration to enhance the algorithm in the further studies.
Keywords:	Deduplication, Rapid Asymmetric Maximum (RAM), CDC, AE, Throughput.
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	DYNAMIC MULTI-PATH ROUTING PROTOCOL HINGED ON FITNESS VALUE USING GA IN MOBILE ADHOC NETWORKS
Author:	G BALU NARASIMHA RAO, ASESH KUMAR TRIPATHY
Abstract:	A MANET is a cluster of different wireless hardware devices connected together temporarily hinged on the frequency in the absence of any framework. Owing to a node's periodic mobility, the network's node layout will regularly alter, creating unstable connectivity and a reduced delivery rate. The Collision of packets in the network is the prime obstacle in the wireless network caused by the node's mobility in different manners with arbitrary speeds that leads to the increase in the collision possibility reflected in overall throughput, and delay between the nodes. Here, we proposed a flexible multipath protocol dependent on the fitness values and genetic algorithm. To choose the optimal path among the multiple paths available to deliver the packets we used the fitness function value (FF). Later we equate our proposed algorithm with HACO-FDRPSO and AOMDV-TC. Performance computation for the proposed algorithm was computed by various parameters such as packet delivery ratio(PDR), throughput (Th), End to end-to-end delay (e2e), and Energy Consumption(Ec).
Keywords:	Unstable Link, Fitness Value, Genetic Algorithm, Collision, Energy, And Flexible
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	A COMPARATIVE STUDY OF CRIME EVENT FORECASTING USING ARIMA VERSUS LSTM MODEL
Author:	M.MUTHAMIZHARASAN, R. PONNUSAMY
Abstract:	This comparative analysis delves into crime event prediction using two distinct methodologies: ARIMA (Auto-Regressive Integrated Moving Average) and LSTM (Long Short-Term Memory) neural networks. Anticipating criminal activities holds immense significance for law enforcement and public safety. ARIMA, a conventional time series forecasting method, relies on historical data patterns to foresee future crimes. In contrast, LSTM, a type of recurrent neural network, excels in capturing intricate, long-term dependencies within data sequences. The study systematically assesses the performance of both models by evaluating their accuracy, efficiency, and versatility in handling diverse datasets. While ARIMA is a reliable choice for fundamental time series forecasting, it encounters challenges in deciphering complex patterns and non-linear relationships in crime data. LSTM, harnessing its deep learning capabilities, demonstrates superiority in capturing these subtleties, resulting in more precise crime predictions. These findings underscore the significance of advanced machine learning techniques, specifically LSTM, in augmenting the accuracy of crime event forecasting. Ultimately, this enhancement aids law enforcement agencies in devising proactive strategies for crime prevention.
Keywords:	ARIMA (Autoregressive Integrated Moving Average), LSTM (Long Short-Term Memory), Crime Prediction, Forecasting, Deep Learning, Non-Linear.
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	HANDWRITTEN DIGIT RECOGNITION USING MACHINE LEARNING
Author:	NIRASE FATHIMA ABUBACKER, MAFAS RAHEEM
Abstract:	The handwritten digit recognition system is a popular research topic, and much research has been done throughout the years. The implementation of this system will be beneficial for many sectors in today’s world. Various types of algorithms can be used to develop a solution for this system. However, the accuracy of the results plays an important role in determining the best solution for the handwritten digit recognition system. In this project, selected machine learning and deep learning algorithms were used to build models to find the most suitable model with the best possible accuracy. According to the results, the CNN model performed better than the other models with an accuracy of 99.25% and 0.99 for each Precision, Recall and F1 Score compared to all the other models.
Keywords:	Digit Recognition, Handwritten, Recognition Model; Machine Learning; Deep Learning
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	MULTI-STRATEGY BASED FUZZY ENHANCED MONARCH BUTTERFLY OPTIMIZATION ALGORITHM (MS-FEMB0A) FOR TASK SCHEDULING AND LOAD BALANCING IN CLOUD ENVIRONMENTS
Author:	DR.SARAVANAN.M.S, MADHAVI KARUMUDI
Abstract:	With the success of cloud computing, enterprises started reducing investments on computing resources by using cloud resources. Outsourced tasks of those enterprises are to be processed in cloud without violating Service Level Agreements (SLAs). Therefore, efficient scheduling of tasks in cloud, in presence of millions of users and diversified workloads across the globe, is NP-hard problem. To solve this problem many bio-inspired optimization algorithms came into existence. One such algorithm is known as Monarch Butterfly Optimization (MBO) which suffers from insufficient exploration power and lack of balance between exploration and exploitation. Moreover, most of the algorithms focused either on scheduling or load balancing though they are related. To address all these problems, in this paper, we proposed an algorithm known as Multi-Strategy based Fuzzy Enhanced Monarch Butterfly Optimization Algorithm (MS-FEMB0A) for efficient task scheduling in cloud leading to load balancing. MS-FEMB0A considers multiple objectives and also multiple strategies such as fuzzy, greedy and self-adaptive towards improving search capability and convergence speed. Simulation study made using CloudSim framework showed that MS-FEMB0A outperforms existing algorithms such as PSO and MOPSO.
Keywords:	Cloud Computing, Load Balancing, Task Scheduling, CloudSim, Multi-Strategy Scheduling
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	THE EFFECTIVENESS OF ONLINE PROBLEM-SOLVING STRATEGIES AND THEIR IMPACT ON THE ACHIEVEMENTS OF ISLAMIC CULTURE IN JORDAN
Author:	MAHMOUD ALSALTI, FADI BANI AHMAD, ABEER MAHMOUD ALRAGGAD
Abstract:	This study examines the effectiveness of online problem-solving strategies and their impact on achievements within the Islamic culture context in Jordan. Investigating the intricate dynamics involved, the study explores the interplay between technology adoption, cultural heritage, and educational outcomes. The study involves 200 participants at Al-Zaytoonah University, delving into technology adoption in education. Utilizing a meticulously designed instrument with both closed and open-ended questions, it evaluates experiences related to ease of use, usefulness, and Cultural heritage, aiming to unravel the interplay between technology and culture. This contributes to discussions on educational progress in Jordan, highlighting the importance of a comprehensive approach to online education. The findings emphasize the need for educators and policymakers to enhance user-friendliness while ensuring cultural attunement and alignment with students' values. In enhancing our understanding, this study provides insights into the optimal integration of online problem-solving strategies for fostering academic success within the framework of Islamic culture in Jordan.
Keywords:	Online Problem-Solving Strategies, Islamic Culture, Technology Acceptance Model, Culture Heritage
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	ENSEMBLE MACHINE LEARNING ALGORITHM METHODS FOR DETECTING THE ATTACKS USING INTRUSION DETECTION SYSTEM
Author:	NAZREEN BANU. A, DR.SKB SANGEETHA
Abstract:	Everyday improvement in distributed computing administrations needs more regard to convey the information with security in light of Interruption happening in a decentralized climate. Cloud security needs headways in the Interruption recognition framework in view of material science and web security. The vast majority of the current IDS checking framework wasn't ready to make the component choice and arrangement really to recognize the interruption. Because of expanding more component dimensionality and non-related highlights drives mistakes to accomplish the presentation. To determine this issue, we propose an Ensemble machine learning method for execution. Hybrid Whale optimization algorithm (WOA) using genetic algorithm and Random Forest Integration is executed. Preprocessing is first done to determine the scaling factor and feature margins. Subsequently, the feature outline is marginalised by estimating the User Behaviour Analysis based on Flow and Time-Based Features. Behavioural Features for Frequency of Protocols is computed to highlight the features based on the variation features. Then, at that point, Hybrid Whale optimization algorithm (WOA) to reduce the number of unrelated highlights, a genetic algorithm is used to choose the highlights. In order to identify the IDS, the selected highlights are finally ready for Random Forest Integration. The proposed framework accomplishes superior execution by distinguishing ensemble methods for IDS to sort the interruption level. Also to listen to the dimensionality idea of future varieties to achieve best location exactness contrasted with different frameworks.
Keywords:	Intrusion Detection System, WOGA, RF, Virtual Machine Monitor, NF, FAR.
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text

Title:	AN EFFICIENT ANOMALY DETECTION BASED HEMODIALYSIS MORTALITY PREDICTION FOR MIXED HEMODIALYSIS DATABASES
Author:	T HEMALATHA , K.V.D KIRAN
Abstract:	As the size of mixed databases increases, challenges arise in improving the true positive rate. Several critical issues such as missing values, attribute noise, and imbalanced classes substantially impact data accuracy. High-quality input data becomes imperative to optimize classification algorithms, especially in the context of imbalanced mixed data types. Thus, the optimization of classic machine learning models becomes essential to ensure accurate predictions on imbalanced datasets. This study addresses these challenges inherent in hemodialysis mixed datasets by proposing an enhanced ensemble classification model incorporating optimal filtering and classification algorithms. A novel framework is proposed to handle missing data with classes, feature ranking, and ensemble classification approaches to improve the true positive rate and error rate on imbalance hemodialysis databases. Experimental results have demonstrated that the proposed approach outperforms conventional techniques in terms of statistical metrics.
Keywords:	Imbalance Hemodialysis Dataset, Probabilistic Classification , Support Vector Machine, Ensemble Learning Model.
Source:	Journal of Theoretical and Applied Information Technology 15^th March 2024 -- Vol. 102. No. 5-- 2024
Full Text