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

Welcome to Journal of Theoretical and Applied Information Technology

Submit Paper / Call for Papers

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

Journal of Theoretical and Applied Information Technology
January 2025 | Vol. 102 No.2

Title:	IMPROVING COLLECTION OF DATA TYPE EVIDENCE AND THE INTEGRITY OF EVIDENCE COLLECTED USING SHA-256 HASHING ALGORITHM FOR WEB BROWSERS
Author:	NUR HAZIERAH MOHD SHAH, AZIAH ASMAWI, SHARIFAH MD YASIN
Abstract:	This study introduces a method to enhance web browser evidence collection in digital forensic investigations. The focus of this study specifically operating 3 evidence collection software forensic toolkits in one developed forensic toolkit called ForenWebSight (FWS). Data is collected from 4 most popular web browsers, Google Chrome, Mozilla Firefox, Microsoft Edge, and Opera in the Windows 11 environment in the context of evidence collection, emphasizing the significance of 35 data types such as history visits, history search, search keyword, cookies, cache, file, session, bookmarks, downloaded files and many more in digital forensic investigations. The existing tools for evidence collection primarily rely on SHA-1 hashing and using older version windows and software toolkits version. Therefore, this study proposes the addition in toolkits implementation, the latest software tools version and the latest solution, an improvement proof-of-concept utilizes SHA-256 hashing algorithm to improve the collection of evidence and enhance integrity. The use of the SHA-256 hash algorithm currently considered secure and resistant to collision attacks. It offers a higher level of security than SHA-1. The evaluation involves comparing the ForenWebSight (FWS) with previous study shows the importance of robust evidence collection tools and methodologies in combating cybercrimes.
Keywords:	Digital Forensics, Web Browsers Evidence Collection, SHA-256 Hashing
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	DEVELOPMENT OF EFFICIENT PROTOCOLS FOR THE SECURE TRANSMISSION OF TRAINING PARAMETERS IN A FEDERATED NETWORK USING ELLIPTIC CURVE CRYPTOGRAPHY
Author:	NARENDRA BABU PAMULA , AJOY KUMAR KHAN ,ARNIDAM SARKAR
Abstract:	Federated learning has become a key concept in collaborative machine learning, allowing several clients to train models independently of one another's raw data. However, federated learning's decentralized structure poses serious security risks, especially when it comes to securely transferring training parameters between clients and the central server. This research proposes the construction of effective protocols employing Elliptic Curve Cryptography (ECC) for the safe transfer of training parameters in a federated network in order to address these issues. ECC is selected because of its robust security characteristics and computational effectiveness, which make it especially appropriate for situations with limited resources, which are frequently found in federated learning scenarios. The suggested approach makes use of ECC to enable safe key exchange, guaranteeing that training parameters are encrypted in transit to thwart manipulation and unwanted access. Furthermore, the protocol is made to reduce communication overhead, which improves the federated learning process's overall effectiveness. A thorough security analysis shows how resistant the protocol is to common risks like eavesdropping and man-in-the-middle assaults. The efficiency of the protocol is further supported by experimental results, which demonstrate notable reductions in computation time and energy usage when compared to conventional cryptography techniques. To sum up, this work opens the door for federated learning to be more widely used in security-sensitive applications by offering a reliable and effective method for safeguarding communication in federated learning settings.
Keywords:	Federated Learning, Elliptic Curve Cryptography (ECC), Secure Transmission, Training Parameters, Cryptographic Protocols, Data Security, Key Exchange, Parameter Aggregation, Communication Efficiency, Network Security
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	INFLUENCE OF DIGITAL TECHNOLOGIES ON INFORMATION SECURITY IN THE PUBLIC ADMINISTRATION SYSTEM
Author:	MYKOLA DURMAN, OLENA DURMAN, DMYTRO DROZHZHYN, IRYNA KRYLOVA, YURIY GAVRYLECHKO
Abstract:	The use of digital technologies in public administration is a relevant issue in the face of the rising number of cyberattacks and threats to information security. The current trend demands operative substantiated decisions for the protection of state authorities. The primary aim of this article is to analyse the influence of digital technologies on information security in public administration systems. The research methodology involved the use of synthesis, generalisation, monitoring, and comparison methods for the evaluation of the current state of cybersecurity in public administration. The work has identified the main directions and means to prevent cyberattacks such as the introduction of cloud solutions, data encryption and the creation of national cybersecurity centres. The study results showed that integration of modern technologies ensures a reduction in the number of successful cyberattacks, enhances the level of information security and facilitates the development of international cooperation. The article deals with the limitations, which governments face while implementing the latest solutions, in particular insufficient financing and staff shortage. The practical significance of the work is in the provision of recommendations for the development of state programs for data protection and enhancing cybersecurity levels in public administration systems. Future studies shall be directed at the implementation of new instruments for threat monitoring and analysis of the effectiveness of international cybersecurity protocols.
Keywords:	Digital Technologies, Information Security, Public Administration, Blockchain, Cybersecurity, Cloud Technologies, International Cooperation
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	SYSTEM ARCHITECTURE OF DIGITAL ASSET MANAGEMENT WITH AI TRISM
Author:	PINYAPHAT TASATANATTAKOOL , PANITA WANNAPIROON, PRACHYANUN NILSOOK
Abstract:	This scholarly investigation seeks to devise the architectural framework of a digital asset management system utilizing Artificial Intelligence-based Trust, Risk, and Security Management (AI TRiSM). The architecture is evaluated through the insight of nine specialists in the domain of information technology, employing five principal inquiries as follows: 1) Data about the characteristics and components of digital asset management 2) Devices & Connections 3) Methodology 4) Technology with AI TRiSM 5) Reports arising from architecture and supporting the university's governance principles. The findings from this evaluative process will be instrumental in advancing a digital asset management platform. The evaluation outcomes are commendable, exhibiting a mean score of 4.53 and a standard deviation (S.D.) of 0.49. In the context of this research, the investigator has meticulously synthesized the technologies employed to ensure that this system is dependable, mitigates risks, and guarantees operational safety. The findings presented in this report facilitate informed decision-making regarding digital asset management, emphasizing the fostering of robust governance within higher education institutions. This approach enhances the efficiency of managing digital resources and aligns with best practices in data security and compliance, ultimately contributing to a more sustainable and accountable educational environment.
Keywords:	Digital Asset Management, Artificial Intelligence, AI-TRiSM
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	INTELLIGENCE WISDOM REPOSITORY MANAGEMENT PLATFORM (I-WRMP) TO PROMOTE LIFELONG LEARNING SKILLS AND LEARNING INNOVATION
Author:	PONGSATON PALEE, JITTIMA PANYAPISIT, THIPWIMOL WANGKAEWHIRAN, CHARNNARONG KAMPHET, ADIREK YAOWONG
Abstract:	This research project the research objectives are as follows: 1) To analyze Systematic reviews and Meta Analyses and the bibliometrics analysis of research related to promoting lifelong learning skills and learning innovation with Intelligence Wisdom Repository Management platform (I-WRMP) 2) To develop a prototype, I-WRMP. 3) To evaluate the I-WRMP use of the in Chachoengsao Province (Thailand) in real situations. Research is an application of the research and development concept (Research and Development). The target group is administrators, teachers, a total of 113 people, and schools under local administrative organizations in Chachoengsao Province. The research framework is divided into 3 phases, Phase 1 Analysis and synthesis of the digital intelligence system model to promote lifelong learning skills based on innovation and A new way of life in Chachoengsao Province. Phase 2 Development of prototype digital intelligence system to promote lifelong learning skills based on innovation and a new way of life in Chachoengsao Province. Phase 3 Evaluation of the use of the digital intelligence system to promote skills. The interdisciplinary nature of the field was underscored by the varied subject areas involved, including Computer Science, Social Sciences, and Engineering. The research concludes that while substantial progress has been made in understanding and implementing (I-WRMP), further exploration is necessary to optimize these systems for practical applications. This continued research is crucial to leverage AI full potential, enhancing lifelong learning skills and supporting continuous, self-directed education across diverse contexts. The evaluation results showed that the developed I-WRMP was the most suitable, with the combined mean of 4.77, and the standard deviation was 0.43 and indicating that the results of the measurement before studying and the learning achievement after studying with the normal teaching method were statistically significantly different at the .01 level.
Keywords:	Intelligence Wisdom Repository, Lifelong Learning skills, Learning Innovation
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	NETWORK DIVERSITY AND GREEN INNOVATION PERFORMANCE IN THE DIGITAL ECONOMY A CHAIN-MEDIATION MODEL OF GREEN PERCEPTION SIMILARITY AND TECHNOLOGICAL INNOVATION
Author:	DONG LIU, SZE-TING CHEN
Abstract:	This inquiry delves into the nuanced relationship between network diversity and green innovation performance within the digital economy's framework, with a particular emphasis on the sequential mediating effects of green perception similarity and technological innovation. Drawing on structural equation modeling and survey data from 679 manufacturing firms in Jiangxi, China, the findings underscore the positive impact of network diversity on green innovation performance, mediated by green perception similarity and technological innovation. Moreover, the study illuminates the pivotal role of green dynamic capabilities, augmented by digital technologies, in harnessing network diversity for enhanced innovation outcomes. This research not only offers strategic insights for businesses aiming to bolster their green innovation capabilities through digital transformation but also introduces a novel perspective on the interplay between network diversity and green innovation performance.
Keywords:	Network Diversity, Green Innovation, Green Perception Similarity, Technological Innovation, Dynamic Capabilities, Digital Transformation
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	CHRONIC KIDNEY DISEASE PREDICTION USING AN ENSEMBLE OF SVM, LR, AND MLP MODELS
Author:	MS. GREESHMA ELIYAN, DR. ARUNADEVI KARUPPASAMY, MS. INDU GOVINDA PILLAI, MS. ANJU MATHEW CHERIYAN
Abstract:	Chronic kidney disease (CKD) is a global issue, and it needs to be detected as early as possible for the effective treatment. Many studies applied Machine learning (ML) techniques for predicting the CKD in early stages. Different approaches were used in building models using ML algorithms. The algorithms were trained individually, and ensemble approach were used to build the model. Different datasets were used for different studies. The accuracy of the model is the main concern in the prediction of diseases. It is important to build a model which can produce the highest accuracy. In this study we aim to build a model using different machine learning approaches to predict the CKD in early stages. The ML algorithms such as K Nearest Neighbor (KNN), Random Forest (RF), Multilayer Perceptron (MLP), Logistic Regression (LR) and Support Vector Machine (SVM) were used to do the experiment. RF and KNN were applied individually on CKD dataset and SVM, MLP and LR were ensembled using Voting Classifier. The performance of the models was analysed in terms of accuracy, precision, F1-score and recall. The ensembled method showed the highest accuracy of 100% compared to other individual models. This proves that the ensemble model performed better than each individual models.
Keywords:	CKD Prediction, Machine Learning, SVM, LR, MLP, Ensemble model, Voting Classifier
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	FROG LEAP INSPIRED OPTIMIZATION-BASED EXTREME LEARNING MACHINE FOR ACCURATE CLASSIFICATION OF LATENT AUTOIMMUNE DIABETES IN ADULTS (LADA)
Author:	B. SUCHITRA, J. RAMKUMAR, R. KARTHIKEYAN
Abstract:	Latent Autoimmune Diabetes in Adults (LADA), also known as type 1.5 diabetes, represents a hybrid form of diabetes with characteristics of both type 1 and type 2 diabetes. Unlike type 1 diabetes, which has a rapid onset, LADA develops gradually, often diagnosed in adults over 30. In contrast to type 2 diabetes, LADA is caused by an autoimmune response that progressively destroys insulin-producing beta cells. This overlap often leads to misdiagnosis, as LADA is commonly mistaken for type 2 diabetes, delaying appropriate treatment. Classification algorithms face challenges in predicting LADA due to overlapping symptoms, high-dimensional health data, and imbalanced datasets. Existing methods lack robustness in accurately differentiating LADA from other diabetes types, leading to frequent misclassifications and treatment inefficiencies. To address these challenges, the Frog Leap Inspired Optimization-based Extreme Learning Machine (FLIO-ELM) has been proposed. FLIO-ELM combines bio-inspired optimization with a single-layer neural network framework. Inspired by frog-leaping behavior, this method enhances input weight and bias optimization, ensuring better feature selection. Frogs in the optimization represent potential solutions, with local and global search phases refining parameters. The Extreme Learning Machine (ELM) framework computes output weights using the least squares method, ensuring fast training and robust generalization. This hybrid mechanism balances exploration and exploitation to improve prediction accuracy. FLIO-ELM achieves significant improvements in performance metrics, with an 81.304% classification accuracy and an 82.093% F-measure. Its false discovery and omission rates are minimized, ensuring reliability in LADA prediction. These results establish FLIO-ELM as an effective diagnostic tool.
Keywords:	LADA, Frog Leap Inspired Optimization, ELM, Diabetes, Bio-Inspired Algorithms
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	ENHANCING CRISIS SOLUTIONS FOR IDPS IN IRAQ THROUGH GIS DATA ANALYSIS USING SVM
Author:	SHAWQ SALMAN AL.KHAFAJI, KIFAH TOUT, ZAID F. MAKKI
Abstract:	Internally displaced persons (IDPs) in Iraq face significant challenges due to multiple reasons including natural disasters, conflicts and inadequate infrastructure. In order to ensure the safety of these population groups and improve their conditions, strategies must be in place to manage these crises effectively. In this study, we proposed an approach that is based primarily on the integration of machine learning represented by Support Vector Machines (SVM) and Geographic Information Systems (GIS) to enhance crisis management solutions for IDPs in Iraq. The information provided by GIS was utilized, analyzed and classified by the SVM classifier in order to predict in advance the areas exposed to crises in order to achieve the ultimate goal of allocating resources and increasing the speed of response. In this study, the analysis of the data resulting from GIS and its classification using the weights affecting the result were discussed to predict the best path taken by the IDPs and to expedite the provision of livelihoods to them to avoid catastrophic consequences and the aggravation of the crisis. The proposed method was proven effective through the results we reached through training and testing to help in decision-making and rapid response.
Keywords:	Machine Learning, Data Analysis, Support Vector Machines (SVM), Geographic Information Systems (GIS), Management of Crisis, Internally Displaced Persons (IDPs), Prediction.
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	KEYSTROKE DYNAMICS ON MULTI-SESSION AND UNCONTROLLED SETTINGS USING CNN BI-LSTM
Author:	SEPTIO RAHMAN PUTRA , ANDRY CHOWANDA
Abstract:	Keystroke dynamics is a behavioral biometric that identifies individuals based on their typing style and rhythm. It is a non-intrusive and low-cost authentication method that does not require additional hardware. This study addresses the challenge of continuous authentication using keystroke data collected from multiple sessions and long periods. A hybrid Convolutional Neural Network (CNN) and Bidirectional Long Short-Term Memory (Bi-LSTM) model was developed to capture both spatial and temporal dependencies in the keystroke data. The research process involved data collection from 10 users over six months, preprocessing to remove irrelevant data, and feature extraction to transform the data into a usable format. The model achieved impressive performance with Equal Error Rates (EERs) ranging from 0.009 to 0.127, demonstrating its effectiveness in continuous authentication scenarios.
Keywords:	Keystroke Dynamics, Biometric Authentication, CNN, Bi-LSTM,RNN, Classification, Performance Evaluation, EER.
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	CUCKOO SEARCH OPTIMIZATION-BASED FEATURE SELECTION FOR PREDICTING AUTISM SPECTRUM DISORDER USING ARTIFICIAL IMMUNE ALGORITHMS
Author:	KEERTHI GUTTIKONDA, DR. G. RAMACHANDRAN, DR. G. V. S. N. R. V. PRASAD
Abstract:	Autism Spectrum Disorder (ASD) is a complex neurological, neuro-developmental disorder that exerts influence on a child's social interactions and communication. Early detection and intervention can improve outcomes, but current screening tools have limitations. To address this challenge, this paper proposes an ASD prediction model based on Cuckoo Search Optimization (CSO) and Artificial Immune Algorithms (AIA). The proposed method is designed to select a feature subset of the most informative features from a high-dimensional dataset for use in predictive models. CSO is a meta-heuristic global search optimization algorithm inspired by the behavior of cuckoos in nature. The algorithm is designed to search for the optimal solution by exploiting the search space. Among various AIAs, Clonal Selection Classification Algorithm (CSCA) is evolved as the efficient algorithm to detect various diseases. The proposed CSO-CSCA model attained 95.85% accuracy and stood as a promising approach for the early detection and intervention of the disorder. The results of the present study promise to improve the accuracy of predictive models and support the development of new screening tools for the early diagnosis of ASD.
Keywords:	Autism Spectrum Disorder, Cuckoo Search Optimization, Artificial Immune Algorithms, Clonal Selection Classification Algorithm, Artificial Immune Recognition System
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	THE IMPACT OF MENTORSHIP PROGRAMS USING VIRTUAL REALITY ON STUDENT PERFORMANCE
Author:	OKSANA IVANYTSKA, OLHA BILYAKOVSKA, MAGDALYNA OPACHKO, HALYNA LOTFI GHAHRODI, NADIYA ZAYACHKIVSKA, TETYANA YAKYMOVYCH
Abstract:	The formation of professional knowledge is possible when students interact with a more experienced mentor. The article aims to determine the effectiveness of virtual reality mentorship programmes and their impact on student performance. The study used analysis methods, surveys, the Mann-Whitney U Test Calculator, Student's coefficient and the Thurstone scale. Implementing the mentoring project was made possible with the help of unique virtual reality applications: Immerse, which offers a personalised approach to learning (M=9.04) and an automated approach to monitoring progress (M=9.0), and Edpuzzle, which provides a personalised approach to learning (M=9.03) and the transmission of signals between mentor and student (M=8.94). The application also provided automatic control over the completion of tasks (M=9.01), which contributed to the understanding of foreign language rules. It was found that using virtual reality in the mentoring process contributed to achieving high results by students – group 1: theoretical knowledge – 92 points, practical knowledge – 93 points, group 2: theoretical knowledge – 95 points, practical knowledge – 91 points. The practical significance of the work is aimed at the possibility of using virtual technologies to implement mentoring programmes. Future research may determine students' professional competence based on the mentoring and traditional approach.
Keywords:	Professional Development, Mentors, Interactive Learning, Virtual Reality, Student Performance, Educational Programs
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	A NOVEL APPROACH TO IMPROVE ROUTING PERFORMANCE IN UNDERWATER CHANNEL MODELLING AND CLUSTER HEAD SELECTION
Author:	PRATHIBA N, NAVEEN I G, SEEMA S
Abstract:	Underwater monitoring and communication are crucial for studying the environment, including seismic waves, underwater robots, and water pollution. Wireless sensor networks (WSNs) are being developed to address the challenges of energy efficiency in underwater wireless communication. Batteries are used to power underwater sensor nodes, which are limited in power resources and difficult to recharge. This requires new protocols for routing, considering the differences between terrestrial and underwater networks. However, underwater communication faces challenges such as attention, noise, and energy consumption. Despite these challenges, underwater wireless communication remains a growing field of research. To overcome these issues, we present a novel approach which considers three different aspects as sink localization, channel modelling and Cluster Head (CH) selection. The localization scheme divides the network into multiple layers and appoints a sink for each layer, channel modelling considers underwater noise, attenuation, SNR, propagation delay for channel modelling. Later, the CH selection scheme is presented which uses residual energy, and distance from sink to select the cluster head. Before finalizing the CH, it appoints a temporary CH which helps to improve the stability of Cluster Head selection mechanism. The comparative analysis shows the significant improvement in the network performance in terms of network throughput, and lifetime of the network for varied nodes and radius of network.
Keywords:	Underwater networks, Cluster head selection, Localization, Channel modelling, Network Lifetime
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	THE ROLE OF AI ADOPTION IN ACHIEVING SUSTAINABLE AUDIT QUALITY
Author:	SHAILA VARELIE TRITAMA, NARENDRA ADWITIYA MAHAPRAJNA, BAMBANG LEO HANDOKO
Abstract:	The audit profession plays a crucial role in ensuring financial transparency, with the adoption of Artificial Intelligence (AI) offering significant potential to enhance audit quality. This study investigates factors influencing AI adoption among auditors in achieving sustainable audit quality using the Unified Theory of Acceptance and Use of Technology (UTAUT2) framework. A quantitative approach was applied, collecting data from 130 auditors in Indonesian public accounting firms via questionnaires. Data analysis utilized Structural Equation Modelling-Partial Least Squares (SEM-PLS) using smartPLS. The findings indicate that Facilitating Conditions and Habit significantly influence AI adoption, while factors such as Performance Expectancy, Effort Expectancy, Social Influence, Hedonic Motivation, and Price Value are statistically insignificant towards AI adoption. Additionally, the adoption of AI significantly impacts sustainable audit quality by improving efficiency, reducing errors, and enhancing reliability. These results highlight the need for infrastructure support and habitual use to drive AI adoption among auditors.
Keywords:	Artificial Intelligence, Sustainable, Audit, Quality, UTAUT2, SEM-PLS
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	DETECTION OF ABNORMAL LUNG SOUNDS USING REMOTE STETHOSCOPE VEST COAT WITH DEEP CONVOLUTION NEURAL NETWORKS
Author:	VISHNU VARDHAN B, DR. M. KALAISELVI GEETHA, DR. G SYAM PRASAD
Abstract:	Much healthcare data is collected from various hospitals to analyze and diagnose specific diseases in the medical field. Diagnosing the specific diseases based on the disease patterns and sample analysis is complex. This paper proposes a novel approach to remotely detecting and analyzing lung sounds using a vest coat stethoscope with deep learning algorithms. The proposed system addresses the limitations of traditional stethoscopes, which require physical contact with the patient and can be difficult to use in remote or noisy environments. The proposed system consists of a vest coat stethoscope equipped with a microphone array to capture lung sounds and a deep learning algorithm to process and analyze the data. The system is designed to be lightweight and portable, making it ideal for use in remote or field settings. The experiments were conducted with volunteers in various settings, including a hospital environment and a noisy construction site. The results showed that the proposed system 2dCNN could accurately detect and classify lung sounds in quiet and noisy environments, with a high level of accuracy compared with the 1D-CNN Model and LSTM. Overall, the proposed system shows great promise for remote diagnosis and monitoring of lung conditions, particularly in settings where traditional stethoscopes may not be effective. Future work will focus on refining the deep learning algorithms and expanding the system to include additional physiological signals for a more comprehensive analysis of lung health.
Keywords:	1D-CNN Model, LSTM, 2D-CNN, Lung Sounds.
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	THE ROLE OF FAMILY FUNCTIONING AND PARENTAL ATTITUDES IN PREDICTING ADOLESCENT ADOPTION OF AI TECHNOLOGY
Author:	NIE NI, ZEINAB ZAREMOHZZABIEH, MANSOUREH ZAREAN, SEYEDALI AHRARI, KHADIJE BARZEGAR
Abstract:	As AI becomes increasingly integrated into adolescents' lives, it is essential to understand the factors that shape their engagement with AI to foster responsible and confident use. This study examines how family dynamics, based on the McMaster Model of Family Functioning (MMFF), influence adolescents' adoption of AI, focusing on family functioning, parental attitudes, and gender. Using SmartPLS 4, we assessed these relationships. Findings indicate that specific dimensions of family functioning, such as communication, affective responsiveness, and affective involvement, significantly predict adolescents’ confidence and interest in AI, with parental attitude further enhancing these effects. Gender differences also emerged, suggesting that boys and girls respond differently to family interactions in their engagement with AI. The findings emphasize the importance of family in shaping responsible AI use and suggest that fostering open communication and emotional support within families can improve adolescents’ digital readiness. This study contributes to the growing literature on family influences in technology adoption, recommending family-centered strategies to equip adolescents with the skills and attitudes required for an AI-driven future. Future research directions include examining cross-cultural variations and the longitudinal impacts of family dynamics on technology adoption patterns in adolescence.
Keywords:	Adolescent AI Adoption, Family Functioning, McMaster FAD, Parental Attitudes, Gender Differences, Problem-solving
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	LEVERAGING ADVANCED DEEP LEARNING TECHNIQUES TO PRIORITIZE KEY FOCUS AREAS IN THE TEXTILE SECTOR FOR PROGRESS TOWARDS INDUSTRY 4.0
Author:	YOUNES JAMOULI, MOUHSENE FRI, AZIZ SOULHI
Abstract:	Companies can increase their efficiency and competitiveness by implementing Industry 4.0 approaches. However, the Moroccan textile and apparel industry severely restricts the use of these technologies. Developing a logical implementation strategy, such as the proper emphasis areas prioritization, remains a contentious issue and a barrier for practitioners despite all the advancements in these theories and practical approaches. This article uses deep learning to successfully integrate the industry 4.0 paradigm and, based on an intelligent model, develops a support tool for the apparel stakeholders. With the help of SIRI Dimensions maturity and the advancement of a set of common key success factors (CSFs), a neural network was trained to forecast the tailoring of Industry 4.0 implementation focus areas priority. These SIRI Dimensions and CFSs were chosen as input data. The neural network model has since been trained, tested, and validated using the dataset. Twenty percent of the data was used to assess the trained network, and a tuning hyperparameter procedure was created to improve the model's performance. Accuracy, precision, and the specified loss function have all been assessed and optimized for performance indices including Categorical Cross Entropy (CCE). With an accuracy of 96.5% for the Organization focus area, 93.9% for the Technology focus area, and 92.8% for the Process focus area, the proposed model may then determine the appropriate priority of focus areas for Industry 4.0 deployment. The findings of this study could serve as a starting point for other researchers who wish to create a roadmap for successful digital transformation by developing initiatives tailored to each priority focus area.
Keywords:	Industry 4.0, Critical Success Factors, Implementation Priority, Neural Network, Textile And Clothing Industry
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	DATASET OF INDUCED FLUORESCENCE SPECTRA FROM HUMAN LIVER BIOPSIES
Author:	JESÚS YALJÁ MONTIEL PÉREZ, CÉSAR CASTREJÓN PERALTA, JONATHAN AXEL CRUZ VAZQUEZ, JOSÉ MANUEL DE LA ROSA VÁZQUEZ, DIEGO ADRIAN FABILA BUSTOS, KAREN ROA TORT, JOSUÉ DANIEL RIVERA FERNÁNDEZ, GALILEO ESCOBEDO
Abstract:	Under the premise of literature identifying the molecules that constitute the liver and their alterations in disease cases, liver biopsy stages can be marked with fluorescence intensities induced by their composition. This article presents a statistical analysis of fluorescence spectra induced in 401 liver biopsies, associated with four disease states previously classified by pathologists using the METAVIR scale. The spectra are presented as a database containing 7,551 intensity measurements. To induce fluorescence in liver tissue, three light sources with wavelengths of 330 nm, 365 nm, and 405 nm were used, wavelengths close to the excitation of the main components for characterizing liver disease stages.
Keywords:	Fluorescence Spectra Of Liver, Induced Fluorescence, Medical Data Analysis, Photonics, Spectrum Dataset.
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	MACHINE LEARNING APPROACHES IN FANTASY CRICKET: COMPARATIVE EVALUATION OF USER-CREATED, RANDOM, AND K-MEANS CLUSTERING METHODS
Author:	POLINATI VINOD BABU , Dr. M.V.P. CHANDRA SEKHARA RAO
Abstract:	This research examines three approaches to building fantasy cricket teams: User-Created Procedure, Random Procedure, and K-Means Clustering Algorithm. The objective is to identify the best formation strategy by examining player performance data from six games. While the Random Procedure creates teams by picking players at random within predetermined parameters, the User-Created Procedure uses manual selection based on intuitive strategies. Using a machine learning technique, the K-Means Clustering Algorithm groups teams according to credit and performance indicators to find the best-performing teams that stay within credit restrictions. This optimizes team formation. Our findings show that, in terms of overall performance, user-created and randomly generated teams are regularly out performed by the K-Means Clustering technique. This study demonstrates how machine learning techniques can improve the development of fantasy cricket teams by providing a data-driven method that is superior to traditional and random approaches.
Keywords:	K-Means Clustering, Fantasy Cricket, Fantasy Points System, Team Optimization, and Player Performance.
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	LEARNING EXPERIENCE TRANSFER SYSTEM ARCHITECTURE WITH ARTIFICIAL INTELLIGENCE ENGINEERING
Author:	SASITORN ISSARO , PRACHYANUN NILSOOK , PANITA WANNAPIROON
Abstract:	Most studies of the learning transfer process involve considering the transfer of academic results or credits from studies between institutions or programs. This is challenging because individual experiences are abstract. Each person's experiences must be transferable, that is to say, concrete, and tangible for use as information, allowing institutions to proceed with the transfer process and raise the level of one's career. Therefore, this article aims to design the architecture of the learning experience transfer system using an artificial intelligence engineering process. Learning experiences occur concerning each person and are transferred to learning experiences using text pre-processing. Text pre-processing consists of six steps: Step 1, Text cleaning; Step 2, Tokenization; Step 3, Lexical analysis; Step 4, Stop words; Step 5, Semantic analysis; and Step 6, Finding the weight of words, i.e., the importance of words (Keyword), using the TF-IDF method to find the weight of each word. Then, similarity values are found to compare the similarities between learning experiences and professional competency levels. Finally, the research takes and stores the learning experience in a knowledge-based manner. The experimental results showed that the Random Forest algorithm had the highest predictive value with Accuracy equal to 100%. This outstanding performance underscores the potential of the Random Forest algorithm in our proposed learning experience transfer system, instilling confidence in its effectiveness and reliability. When the system architecture was designed and evaluated by experts, the results showed the highest level of appropriateness.
Keywords:	Learning experience, Artificial intelligence engineering, Machine learning, Text pre-processing
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	INTEGRATING DRONE TECHNOLOGY AND MACHINE LEARNING FOR ENHANCED FLOOD RISK PREDICTION
Author:	ASEP ID HADIANA , HANDOKO RUSIANA ISKANDAR , RIDWAN ILYAS
Abstract:	This study investigates the integration of high-resolution multispectral and topographic data obtained through drone technology with machine learning to enhance flood risk prediction. Using a multispectral GeoTIFF file covering a designated flood-prone area, critical feature such as the Normalized Difference Vegetation Index (NDVI), slope, and Terrain Ruggedness Index (TRI) were extracted to train a logistic regression model. The model achieved an accuracy of 86.35% and an ROC-AUC score of 0.98, demonstrating strong predictive performance in distinguishing flood-prone from non-flood-prone areas. Feature importance analysis identified low NDVI and high terrain ruggedness as significant predictors of increased flood susceptibility. Despite its strengths, the model showed a tendency to overpredict flood risk, resulting in a higher false-positive rate. This highlights the need for further refinement, including the incorporation of additional data sources such as historical flood records and rainfall data, as well as the exploration of advanced machine learning models to improve precision and reliability. Overall, this study demonstrates the potential of integrating drone-derived data with machine learning for flood risk management. The proposed approach offers a scalable solution for real-time flood prediction, providing actionable insights for improving disaster preparedness and response in flood-prone regions.
Keywords:	Flood Risk Prediction, Machine Learning, Disaster Management, Drone Technology, Multispectral Data
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	UNLOCKING THE POWER OF TRAFFIC SIGN RECOGNITION –AN INTELLIGENT SYSTEM
Author:	N. Sudhakar Reddy, Panthangi Venkateswara Rao, Dr. P. Punitha, Venkata Raghu Veeramachaneni, A. S. Malleswari, Anjaneyulu Gurram, Chetla Chandra Mohan
Abstract:	Intelligent transportation systems have become increasingly important in recent years as a means of improving road safety and traffic flow. One key component of these systems is traffic sign recognition, which enables vehicles to automatically detect and interpret traffic signs in real-time. In this paper, we explore the use of advanced machine learning techniques, specifically Convolutional Neural Networks (CNN) and Decision Tree algorithms, for traffic sign recognition. We propose a system that consists of a camera-based traffic sign detection module, a CNN-based recognition module, and a user interface for displaying the detected traffic signs. The system is trained and tested on a large dataset of traffic signs, and we evaluate its performance using a range of metrics, including accuracy, precision, and recall. Our results demonstrate that the proposed system accuracy. We also investigate the impact of various factors, including lighting conditions, weather conditions, and the distance between the camera and the traffic sign. Our results indicate that the proposed system is robust to changes in lighting and weather conditions, and can accurately detect and recognize traffic signs from a range of distances. Overall, this study provides valuable insights into the use of advanced machine learning techniques for intelligent transportation systems, specifically for traffic sign recognition. The proposed system has the potential to significantly improve road safety and traffic flow, and our findings suggest that future research in this area could lead to even more accurate and reliable systems.
Keywords:	Intelligent, Traffic, Sign, Recognition, CNN
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	RELIABLE UNDERWATER IMAGE CLASSIFICATION WITH ENHANCED CNN MODELS USING QUOKKA OPTIMIZATION
Author:	SARAVANAN P , VADIVAZHAGAN K
Abstract:	This article presents an innovative deep-learning model, QuokkaNet, for underwater image classification tasks. By optimizing the Enhanced CNN (N-CNN) model using Quokka Optimization (QO), QuokkaNet demonstrates significant advancements in performance. The N-CNN model incorporates the layers such as convolutional, pooling, normalization, fully connected, and Nesterov Accelerated Gradient, creating a robust framework for image classification. Quokka Optimization further enhances this model through systematic exploration, exploitation, fitness evaluation, selection, adaptation, and migration. The study highlights QuokkaNet's superior capabilities and effectiveness in underwater image classification compared to DeepSeaNet and MCANet. These findings confirm QuokkaNet's potential as a reliable and accurate underwater research and exploration tool, offering significant improvements in the classification of complex underwater images. The innovative approach and optimization techniques employed in QuokkaNet set a new benchmark for performance in this domain.
Keywords:	Underwater Image Classification, CNN, QuokkaNet, Nesterov Accelerated Gradient, Quokka Optimization, Image Analysis, Classification Accuracy, N-CNN
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	A VIRTUAL EXPERIENTIAL LEARNING PLATFORM THROUGH INTELLIGENT CO-WORKING SPACES TO PROMOTE ENTREPRENEURSHIP AND HAPPINESS LEARNING
Author:	POLPRACHA MONRATTANACHAI, PRACHYANUN NILSOOK, PANITA WANNAPIROON
Abstract:	This research aims to design an information system architecture for a virtual experiential learning platform through intelligent co-working spaces to promote entrepreneurship and happiness in learning. The study employs a qualitative research methodology, encompassing literature review, requirements analysis through interviews and focus groups, detailed architecture design, and expert evaluation. The research presents a three-tiered architecture comprising Front-end Layer, Middle Layer, and Back-end Layer, integrating cutting-edge technologies such as Virtual Reality (VR), Augmented Reality (AR), Artificial Intelligence (AI), and Internet of Things (IoT) with learning theories and entrepreneurial skill development. The proposed architecture focuses on creating efficient learning experiences, adapting to learner needs, fostering collaboration, and prioritizing learner happiness. Expert evaluation indicates that the designed architecture is feasible for real-world implementation and has the potential to revolutionize education and entrepreneurship development. However, challenges remain in data security and integration with existing educational systems. This research provides recommendations for architecture implementation and directions for future research to develop educational systems that meet the needs of 21st-century learners. The findings have implications for educational institutions, policymakers, and technology developers seeking to create more effective and engaging learning environments that prepare students for the challenges and opportunities of the modern world.
Keywords:	Virtual Experiential Learning, Intelligent Co-working Spaces, Entrepreneurship, Happiness Learning, Artificial Intelligence, Virtual Reality, Augmented Reality
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	THE PREDICTION OF SHARIA STOCK PRICE BY USING SUPERVISED MACHINE LEARNING
Author:	FATMA AGUS SETYANINGSIH , SRI ANDAYANI , RETNO SUBEKTI
Abstract:	Investment in Islamic stock assets is currently increasingly in demand by the public. Similar to investing in stocks, this particular investment involves a substantial amount of risk because of the potential for rapid price fluctuations. Consequently, a forecasting tool is required to assist investors in thinking twice before acquiring Islamic stock. The data used is daily data from 2019 to 2022 with a total data of around 1200. The machine learning approaches we selected are variants of the Recurrent Neural Network model, namely Elman recurrent neural network (ERNN), long short-term memory (LSTM), and gated recurrent unit (GRU). The results of GRU models using mean absolute error (MAE) value is 0.0203. The root mean square error (RMSE)is 0.0325 in the GRU model with the best combination of hyperparameters. The model can make prediction values with small error values based on this combination of a proportion of 70% training data and 30% test data. This study recommends that the model is better to use more variations in hidden neurons, layers, activation functions, training algorithms, and parameters to get a better model architecture.
Keywords:	Supervised Machine Learning, Prediction, Sharia Stock
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	JAGUAR-BASED ROUTING PROTOCOL (JRP) FOR IMPROVED RELIABILITY AND REDUCED PACKET LOSS IN DRONE AD-HOC NETWORKS (DANET)
Author:	Dr J RAMKUMAR, VARUN B, V. VALARMATHI, D R MEDHUNHASHINI, R. KARTHIKEYAN
Abstract:	Drone Ad hoc Networks (DANETs) represent an innovative paradigm in wireless communications, utilizing the inherent mobility of drones to create dynamic, self-organizing networks without relying on pre-established infrastructure. Routing within these networks poses significant challenges due to the high mobility of drones, which often disrupts communication links, leading to considerable packet loss and impacting network reliability and performance. To effectively address these challenges, this paper proposes the Jaguar-based Routing Protocol (JRP). Inspired by the stealth and strategic agility of jaguars, JRP is engineered to enhance routing dynamics through several meticulously designed phases. It begins with a scanning phase, where drones assess the network topology and identify optimal communication paths. This is followed by a target selection phase, where the most stable and efficient routes are chosen based on real-time network conditions. The protocol also incorporates a secure data exchange mechanism to safeguard communications against potential security threats. The performance of JRP is rigorously tested through simulations that demonstrate its ability to significantly reduce packet loss and improve energy efficiency. These attributes make JRP a superior choice compared to existing state-of-the-art routing protocols, particularly in environments demanding high network resilience and operational efficiency.
Keywords:	Drone Ad Hoc Networks, Jaguar-Based Routing Protocol, Dynamic Routing, Packet Loss Mitigation, Network Reliability
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	DEVELOPMENT OF A NEW ENCODING ALGORITHM USING VIRTUAL KEYPAD LETTER SUBSTITUTIONS FOR IMPROVED TEXT CLASSIFICATION
Author:	BOUMEDYEN SHANNAQ
Abstract:	A new approach of encoding with help of Virtual Keypad Letter Substitutions has shown in the improved result of text classification. In this study, we focus on a text document classification dataset comprising 2225 documents distributed across five categories: Politics, sports technology entertainment business. However, using both of these vectors, we initially performed traditional machine learning models like Naive Bayes, Logistic Regression, SVM, and Random Forest over the dataset, which provided us with reasonable accuracy, precision, recall, and F1-Score. However, it is hypothesized that the proposed approach, which uses the encoding technique, Virtual Keypad Letter Substitutions, would improve the performance of these models. The encoding method simply converts the letters in the text data with symbols imprinted on a virtual keypad to enhance abstraction that might better capture such features of the text as semantically and syntactically. These findings attest that the models we propose exhibit massive enhancements in all the metrics under study when trained with encoded data. For example, in Naive Bayes, after encoding the datasets into new features, they recorded an accuracy of 95.14%, precision 95.16%, recall 95.14% and F1-score of 95.12% excluding, it revealed inferior performance to that of raw data. The same effects were observed in other models like: Logistic Regression, SVM, Random Forests; Their accuracies were increased by 28,5% to 41,8%. Based on these findings, the authors recommend the Virtual Keypad Letter Substitution encoding algorithm not only as a tool for increasing the accuracy of text classification but also as a tool for data preprocessing in general machine learning. This method is expected to be advantageous in situations where text data comprises of associated formats or noisy data as the encoding may assist in filtering the most appropriate feature for classification. This work provides helpful information for enhancing the dependent variable associated with each type of the predetermined ML model, including C-SVM and naive-bayes for document classification, although its findings are promising for various disciplines, including NLP, Information Retrieval, and Document Classification, where efficient and accurate text classification is crucial for data-driven decision-making.
Keywords:	Text Classification, Virtual Keypad Encoding, Machine Learning, Support Vector Machine, Naive Bayes, Logistic Regression, Random Forest
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	AUTHENTICATION USING FACIAL EXPRESSION DETECTION
Author:	V VIJAYA KUMAR RAJU, NAREGALKAR AKSHAYKUMAR RANGNATH, T BALAJI, DR. T. V. HYMA LAKSHMI, VENKATESWARARAO CHEEKATI, S SINDHURA,
Abstract:	An innovative mixed biometric-based validation model is presented in the paper. Currently, a single biometric check system's recognition accuracy is frequently significantly reduced due to a variety of factors, including the environment, the client's behaviour, and an individual's physiological flaws. Static biometric enrolment is obviously quite vulnerable to pantomime attacks. We suggested crossbreeding two biometric ascribes that consist of physiological and social characteristics because, in practice, a single biometric confirmation only provides one variable of check. The static and dynamic features of a human face are used in this review. Face identification and photo pre-handling are the two key advancements made in order to eliminate a face's important highlights. Naturally, the first step in determining whether a client is authentic or fraudulent is to use facial recognition to assess the client's personality. When at least two similar facial features could result in a generally high match score, it is possible to generate false recognition by solely depending on one modular biometric. However, the rate of misleading dismissal is 11%, whereas the rate of bogus acknowledgment is 0.65%. A certified client will select a look from the seven widely used options previously chosen in the data set using a combination technique that we proposed due to the security flaws in the mentioned circumstance. The chosen look will serve as a secret word to be indisputably identified as a certified or fraudulent client, as evidenced by our results, even when at least two clients happen to have similar faces.
Keywords:	Authentication, Identification, Face, Evidence
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	ENHANCING LOW-RESOLUTION IMAGES THROUGH NOISE FILTERING AND FEATURE PRESERVATION
Author:	HANAN ALI ALRIKABI, DR. MAHEYZAH MD SIRAJ, AHMED MUQDAD ALNASRALLAH
Abstract:	This paper presents a comprehensive resolution enhancement model to increase the quality and make low-resolution images more detailed and clearer. As a result of, images will appear better visually and any lost or obscured details in the original images will be restored. In fields like computer vision and image processing, it can be challenging to enhance the resolution of low-quality images. However, doing so can make it possible to analyze, interpret, and deploy visual data more effectively to be able to accurately analyze or identify significant features, low-resolution images lack the clarity and detail necessary. Through the process of image enhancement, specialists, researchers, and analysts can find hidden patterns or anomalies, extract more pertinent information, and use enhanced visual data to make better decisions. The proposed model's tasks involve combining steps for image enhancement, filtering, and pre-processing. To prepare the data for post-processing, the Labeled Faces in the Wild (LFW) dataset is loaded and resized during the preprocessing phase. Gaussian and Laplacian filtering are two kinds of filtering techniques that are used to enhance the appearance of images, detect edges, and decrease noise. Laplacian filtering assists in edge detection and feature extraction, while Gaussian filtering reduces noise and maintains image details. In the enhancement stage, the filtered image and the original low-resolution image are combined using image blending techniques to produce an enhanced detail level, sharper edges, and better image quality. metrics SSIM and PSNR are used to assess how effective the proposed model is. The results showed that, across a range of image sizes, the proposed model consistently performs better than recent studies, indicating higher performance in enhancing image quality while maintaining structural information. Furthermore, histograms of various sized images provide clarity on how resizing impacts the distribution of pixel density and general image quality and clarity of the image. The proposed model provides potential for use in computer vision and image processing applications and shows significant progress in low-resolution enhancement of images.
Keywords:	Low-Resolution Image, Image Enhancement, Image quality, ML, Gaussian, Laplacian, LFW
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	PREDICTING INSTRUCTOR PERFORMANCE IN HIGHER EDUCATION USING STACKING AND VOTING ENSEMBLE TECHNIQUES
Author:	ABDELBASET R. ALMASRI, NOR ADNAN YAHAYA, SAMY S. ABU-NASER
Abstract:	Instructors' evaluation is crucial to maintaining educational quality and meeting student needs. It is done through a Student Evaluation of Teaching (SET) survey in higher education to provide constructive student opinions to their instructors and help them improve their courses and teaching practices. This study used extensive mining analysis to analyze the students' responses. A public dataset of 5820 SET survey records from UCI was analyzed to reveal insights into the students' perceptions and expectations of how the courses prepare and help them solve real-world issues. In this analysis, the study used six different machine learning methods: K-nearest neighbor (KNN), Support Vector Machine (SVM), Decision Tree (DT), Gradient Boosting (GB), Random Forest (RF), and Extra Trees (ET). The study validated each of these methods individually and in various combinations using two ensemble methods: stacking and voting. The study goal was to identify the best-performing individual methods and determine the best combinations of methods for predicting outcomes. Based on the study, it was found that an ensemble classifier, comprising the four best-performing classifiers (ET, RF, DT, GB) with stacking, performed better compared to other classifiers. This ensemble achieved an accuracy of 91.616%, which was 0.791% higher than the accuracy of the best single-based classifier (ET), which was 90.825%. The results obtained suggest that the use of ensemble learning can effectively enhance instructor performance predictability.
Keywords:	Machine Learning, Instructor, Performance, Prediction, High Education
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	TRANSFORMING IMAGES TO INSIGHTS: OCR-DRIVEN SENTIMENT ANALYSIS FOR MEDICAL DATA CLASSIFICATION
Author:	G DATTA SAI SREYA, P.RAJARAJESWARI, S.HRUSHIKESAVA RAJU
Abstract:	The challenges of existing methods in the image to the text are time-consuming, lack of quality of images, and bringing to the format that is ready for analysis. To overcome these, and to provide better performance, Logistic regression, and Support Vector Machines are applied along with data augmentation for more accuracy. Optical Character Recognition is used to convert the written image into text format, in the review of all the existing methodologies over data augmentation of sentiment analysis over classifying the input data as medical (or ) non-medical classification would be time-consuming and intensive to differentiate and understand the real context behind the data. We conducted a huge analysis of Medical and non-medical data during run-time our model will input the high-resolution image, and convert it to a text file from the input image, which is thereby processed by our proposed model to identify the accurate and fine-tuned algorithm. We have highlighted the significant approaches in NLP (Natural Language Processing). This classification between Medical and Non-Medical data provided prominent results with several pre-trained datasets, which resulted in fetching real-time information from medically approved web information.
Keywords:	Optical Character Recognition (OCR), Tesseract, Natural Language Processing (NLP), Logistic Regression, Support Vector Machine, Generative AI.
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	DIGITAL SOCIAL CAPITAL ON CAREER SUCCESS IN DIGITAL NATIVES AND DIGITAL IMMIGRANT EMPLOYEES
Author:	ANANG KISTYANTO, MUHAMMAD FAJAR WAHYUDI RAHMAN, BUDIONO BUDIONO, NURUL INDAWATI, BIMA YATNA ANUGERAH RAMADHANI, INA USWATUN NIHAYA
Abstract:	This research aims to explore the implications of the industrial revolution 4.0 on the development of the concept of social capital (digital social capital) on employee career success in Indonesia. This study uses a quantitative approach. The research hypothesis was tested using Partial Least Square-Structural Equation Modeling (PLS-SEM) with multi-group analysis (MGA) using Smart-PLS 4. The sample was 215 respondents (101 digital native groups and 114 digital immigrant groups) in Indonesia. The research results found that digital social trust had a positive effect on the career success of digital natives and digital immigrants. Digital social cliques have a positive effect on the career success of digital natives. However, digital social cliques do not affect the career success of digital immigrants. Digital social networking does not affect the career success of digital natives and digital immigrants. Digital social obligation does not affect the career success of digital natives. However, digital social obligation has a positive effect on the career success of digital immigrants. Career success can be studied with a broader interpretation based on digital social trust, digital social cliques, digital social networks and digital social obligations.
Keywords:	Digital social capital, Digital natives, Digital immigrant, Career success, Indonesia
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text

Title:	UTILIZING WORD EMBEDDING’S FOR AUTOMATED QUERY EXPANSION IN ARABIC INFORMATION RETRIEVAL: A BLENDED METHODOLOGY
Author:	YASIR HADI FARHAN, BOUMEDYEN SHANNAQ, SAID AL MAQBALI, OUALID ALI, BASEL BANI-ISMAIL, MUSTAFA TAREQ ABD, MOHANAAD SHAKIR
Abstract:	Search engines face a critical challenge in addressing query-document vocabulary mismatch during Information Retrieval (IR), when user queries do not match the document content. Automatic Query Expansion (AQE) has been widely used to mitigate this issue by identifying related terms. This research work presents a new hybrid AQE technique which includes DMNs within the BM25 model and two techniques, namely EQE1 & V2Q including DAN. It can be asserted that the hybrid technique is characterized by the optimality of the retrieval performances of the two networks, with reduced query drift. Experimental evaluation reveals that EQE1+(DANs+DMNs) obtained P@10 of 44.20LDilde MAP of 33.10 for the TREC 2001, whereas V2Q+(DANs+DMNs) obtained MAP of 30.40 and P@10 of 39.30. However the proposed method BM25+DMNs achieved the highest average MAP of 38.10 for TREC 2001 surpassing all the methods presented in this study. However, it is suggested that additional improvements employing the enhanced embeddings or fine-tuning of the hybrid solutions be implemented due to the drawbacks in expanding the query and positioning of vectors.
Keywords:	Automatic Query Expansion, Information Retrieval, Word Embedding, Deep Averaging Networks, Deep Median Networks
Source:	Journal of Theoretical and Applied Information Technology 31^st January 2025 -- Vol. 103. No. 2-- 2025
Full Text