		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

May 2023 | Vol. 101 No.9

Title:	FPGA AND MACHINING PROCESS - A REVIEW ON PERFORMANCE AND APPLICATIONS
Author:	A.A. SOMANI, R. D. KOKATE , ABHILASHA MISHRA
Abstract:	Machining procedures should deal with contortions of work pieces, machine tools and cutters, that are created by chatter vibration, thermal effect and cutting force. Excellent performance of these sophisticated procedures relies on information availability regarding process conditions. There is a need for dependable systems that integrate signal processing techniques for gleaning useful information. These are useful in the precise machining process, which requires the tightest of tolerances. Machine performance control and prediction become particularly challenging in such conditions of extreme change. Thus, for identifying the actual tool health, a process monitoring or Tool Condition Monitoring (TCM) system is required for various machining processes. Knowledge of tool state or information regarding its condition during machining operations aids in taking relevant actions for effectively performing machining tasks. This paper explores the studies on TCM and investigates the diverse parameters and techniques exploited for TCM in distinct machining operations. This article provides the necessity and significance of FPGA-based monitoring mechanisms for TCM. It investigates the studies on FPGA-based performance monitoring systems utilized for diverse tasks and applications and elucidates the role of FPGA-based performance monitoring systems for developing better and effective future TCM systems useful in diverse machining tasks.
Keywords:	FPGA, Machining Process, Performance Monitoring, TCM, Tool Wear.
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	A COMPARATIVE ANALYSIS ON THE COMBINED MULTI LEVEL FUNCTIONALITY FRAMEWORK IN CLOUD ENVIRONMENT WITH ENHANCED DATA SECURITY LEVELS FOR PRIVACY PRESERVATION
Author:	K.SANTHI SRI , N. VEERANJANEYULU
Abstract:	Cloud Computing (CC) refers to a network of remote servers and user access via the Internet. Distributed data centers all around the globe are responsible for providing the infrastructure and hosting the servers that power cloud services. The dynamic user group handling and securing their sensitive information using cryptography model is a challenging task as the user groups are continuously increasing. Encryption is a better answer for these kinds of problems, but allowing access to users in the cloud has its own set of challenges. From the client's point of view, when the information is saved in the cloud, it should be crypted well to ensure that no other user can read it if they get access to it in any way. Using cloud storage comes with a number of potential drawbacks, including the lack of security for critical information. This study examines the considerations that should be made when choosing a cloud service provider, with a focus on the client's encryption needs and how, without them, the client runs the risk of either losing data or paying more than necessary to the cloud service provider. To enhance the privacy preservation and key management issues a combined framework that handles data encryption, key management and distribution, cloud user group management is analyzed in this research that enhance the data security levels using cryptography model and the key management and distribution process with the dynamic cloud user groups. This research performs a comparative analysis by comparing the combined framework with traditional models. The proposed model when compared with the existing methods exhibit better security levels.
Keywords:	Cloud Computing, Privacy Preservation, Cryptography, User Groups, Data Security, Encryption, Quality of Service.
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	MOBILITY PREDICTION-BASED SOURCE ANONYMITY ROUTING PROTOCOL (MPSARP) FOR SOURCE LOCATION PRIVACY USING NS2 TECHNIQUES
Author:	CHINNU MARY GEORGE, DIVYA SHARMA, REEJA S R
Abstract:	According to information and technology, the three most crucial criteria for success are the three most essential factors in safeguarding privacy in the twenty-first century. As a result, should keep a close check on all of the vast organizations that are watching — companies like Google, Twitter, Facebook, AT&T, and Verizon — making sure they aren't spying on. Every one of these businesses has recently added location-based services to their product or service offerings (or is doing so). Even though I don't reside in a vast geographical area, Twitter can now know what city I'm in and what neighborhood I'm in. Today, consumers face the reality of walking about with a beacon that constantly transmits information about their location to a central server. While capable of broadcasting exact details of our places and movements, cell phones are not the only gadgets that can do so.To address these problems, the study suggested MPSARP ensure location privacy across a wireless network (Mobility prediction-based source anonymity routing protocol). Our proposal is for an Unspecified, efficient routing and location-based system that offers good anonymity protection at a reasonable cost while maintaining excellent performance (MPSARP). By dynamically partitioning a network arena into diverse zones and arbitrarily picking nodes inside zones to serve as intermediary relay nodes, this approach provides a nontraceable unspecified route that is not traceable by the user, resulting in a nontraceable unknown way that is not traceable by the user. The process of simulation is accomplished using NS2 and the performance outperforms the existing techniques.
Keywords:	Mobility, Source Location Privacy, Notify,Piggyback,Routing, Packet Deliver Ratio, Speed, Flows, Delay, Unknown Neighboring Balance.
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	EMOTION RECOGNITION IN ARABIC: A BERT-BASED TRANSFER LEARNING APPROACH LEVERAGING SEMANTIC INFORMATION OF ONLINE COMMENTS
Author:	MAHA JARALLAH ALTHOBAITI
Abstract:	With a wide range of practical applications, such as the diagnosis of mental health disorders and the detection of suspicious online behavior, the recognition of emotions in textual data is a crucial task. However, Arabic emotion recognition remains under-addressed in comparison to other languages, primarily due to the scarcity of labeled data. Pre-trained language models and transfer learning offer promising techniques to overcome the scarcity of labeled data for downstream tasks, such as emotion recognition. In this paper, we comprehensively explore the utilization of pre-trained Bidirectional Encoder Representations from Transformers (BERT) models for Arabic fine-grained emotion recognition. We further propose a straightforward yet effective method for enhancing transfer learning in emotion recognition through the integration of semantic information, specifically, information extracted based on sentiment analysis and named entity recognition. To evaluate our proposed method, we conduct experiments on an existing benchmark dataset using pre-trained BERT models. The results indicate that our proposed approach of integrating the named entity information with the BERT model yielded a high weighted-average F1 score of 81.60%. Compared to the existing studies on Arabic emotion recognition in the literature, our proposed method outperforms the state-of-the-art approach, yielding of 9.80%, 9.72%, and 9.60% in weighted-average F1 score, precision, and recall respectively.
Keywords:	Natural Language Processing, Emotion Recognition, BERT, Pre-trained Language Models, Semantic Information.
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	A FEATURE RECOGNITION SYSTEM FROM STEP AP203 NEUTRAL FILE, BASED ON STEP AP224 MANUFACTURING FEATURES
Author:	OUSSAMA JAIDER, NIHAD AGHBALOU
Abstract:	Geometric modeling has been used for many years as a design tool for CAD (Computer Aided Design) systems. However, data definitions provided by these systems cannot be used directly by a computer-aided process planning system. As a result, the concept of Automatic Feature Recognition (AFR) has been introduced to integrate CAD and downstream applications such as CAPP (Computer Aided Process Planning). However, even though a feature recognition system is capable of recognizing and extracting manufacturing features from a neutral file such as STEP AP203, these features are still defined by geometric and topological low-level data and thus, lack of standardization. For this reason, STEP AP224 application protocol has been developed to provide a standard for both implicit and explicit representations for machining features. Nevertheless, recognizing machining features of AP224 from a neutral file such as STEP AP203 is still the subject of several research. The main purpose of this paper is to explain the methodology of converting manufacturing features obtained by a recognition system we have developed previously, to machining features defined according to ISO 14649. The EXPRESS modelling, Python language and the STEPCode library are the main tools used to map parameters of objects representing manufacturing features obtained by the AFR system, to geometric and dimensional attributes of classes representing features of ISO 14649. An example part was tested to show the results of the developed application, called CAPP-Turn.
Keywords:	AFR, STEP AP224, ISO 14649, CAD/CAPP/CAM
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	PERFORMANCE COMPARISON OF DECISION TREE, RANDOM FOREST, AND XGBOOST MODELS; AND ITS INTERPRETABILITY USING SHAP FOR RECOGNIZING THE NECESSITY OF CAESAREANS SECTION OF CHILDBIRTH
Author:	ASWAN SUPRIYADI SUNGE, EDI ABDURACHMAN, YAYA HERYADI, IMAN H. KARTOWISATRO
Abstract:	Birth is one of the processes that every woman will go through after pregnancy, many risks are faced from birth failure to death of the baby and mother. One of the methods for reducing this risk is by Caesarean section, but before doing this, medical personnel or obstetricians need medical records from the patient before making the decision, from this analysis Machine Learning (ML) is needed to analyze whether Caesar is needed or not for the candidate. mother to give birth. The data used in analyzing the prediction of Caesarean birth with external data which consists of 32 attributes totaling 3602 are then tested with the Decision Tree, Random Forest, and XGBoost models then compared which one has the highest performance of each model with the overall data and selected data then tested with the model SHAP to see which features are highest in Caesar's predictions. The test results using all the features show that Random Forest (0.86) achieves the highest accuracy. However, using selected features such as Age, Duration until the next pregnancy in days, and Obesity has the highest accuracy, namely XGBoost (0.85). Finally, in the SHAP model, the highest feature of the 32 features is ICD10O82 (Contractions but without any indication) and ICD10O80 (Infection of the female reproductive organs). From the results of testing with the ML model and it is also seen that the very domain features are used as a guideline for cesarean prediction decision-makers.
Keywords:	Prediction, Caesar, Classification, Comparison, SHAP.
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	SYSTEMATIC LITERATURE REVIEW ON IMPLEMENTATION OF WHISTLEBLOWING SYSTEM IN PREVENTING FINANCIAL ACCOUNTING FRAUD
Author:	MEIRYANI, DANIEL, SHI MING HUANG, ASL LINDAWATI, DIANKA WAHYUNINGTIAS, AGUNG PURNOMO, AGUSTINUS WINOTO, MOCHAMMAD FAHLEVI
Abstract:	The whistleblowing system is a part of internal control that has not been widely discussed in accounting research in Indonesia. This study aims to determine the influence of the factors that influence the implementation of the whistleblowing system in supporting the prevention of financial/accounting fraud in the corporate environment. In addition, this study wants to examine whistleblowing policy as a moderating variable on the relationship between the factors that influence the whistleblowing system and the implementation of good corporate governance and their relationship with the audit committee. This study uses a qualitative research approach using a systematic literature review to see if there is a pattern of relationship between the factors that influence the effectiveness of the whistleblowing system reporting and good corporate governance. The implication of this research is that companies need to ensure the safety of whistleblowers in the whistleblowing system.
Keywords:	Whistleblowing System, Internal Control, Good Corporate Governance, Reporting Fraud
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	PADDY GRAIN MATURITY ESTIMATION USING DEEP LEARNING APPROACH- YOLOMATURITY
Author:	JAY PRAKASH SINGH, CHITTARANJAN PRADHAN
Abstract:	The detection of crop maturity is one of the most crucial parts of precision farming. Crop maturity estimation can support farmers to harvest their crop at a proper time. Crop harvesting in premature/post mature can lead to crop quality degradation. We have proposed an approach using deep learning that can automate the process of paddy grain maturity detection. Our proposed approach YOLOMaturity (YOLOResNet70Combined) has performed well on available dataset. We have compared our approach with the modifications of the YOLO algorithms, and it outperformed other discussed approaches. Proposed approach yield 10.3 %, 9.6 %, 12.4%, 12.1%, 8.4%, 6.5% and 9.3% better mAP(Mean Average Precision) compared to YOLOv4RCombined, YOLOv4FPN, YOLOv5sFPN, YOLOv5sR, YOLOResNet43HardSwish,YOLOResNet43Combined and YOLOResNet70HardSwish respectively. Proposed approach furnished validation accuracy 97.7 %, training accuracy 97%, validation loss 7.6% and training loss 10%. These results indicate that our proposed approach have promising result and can be utilized for automatic real-time paddy grain maturity.
Keywords:	Precision Farming, Deep Learning, Maturity Detection, Paddy Grain, YOLOResNet70
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	FORECASTING SOLAR PHOTOVOLTAIC ENERGY PRODUCTION USING LINEAR REGRESSION-BASED TECHNIQUES
Author:	SHEREEN A. EL-AAL, MAHA ABDULKARIM ALQABLI, AMANY A. NAIM
Abstract:	Photovoltaic (PV) systems are now viewed as being crucial to the advancement of renewable energy. In order to maximize the performance of the power grid in accordance with market demands and prevent issues with solar generation due to instability, smooth solar energy generation requires precise and trustworthy forecasts. In order to estimate the solar system's output power, machine learning regression models were used to assess the solar system's performance. For the solar generation dataset, machine learning models based on linear regression, such as Multiple, Ridge, Lasso, Decision Tree, and Polynomial regression, have been applied. Performance measures were used to quantify how well the behavior of the suggested technique fit the dataset. The study's findings demonstrate a 93.7% accuracy rate for the polynomial regression model with a lower value of Mean Square Error (MSE).
Keywords:	Photovoltaic, Solar Energy, Machine Learning, Regression, Polynomial Regression.
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	APPROACH TO ATTRIBUTED FEATURE MODELING FOR REQUIREMENTS ELICITATION IN SCRUM AGILE DEVELOPMENT
Author:	KARAM IGNAIM, SULTAN M. AL KHATIB, KHALID ALKHARABSHEH, JOÃO M. FERNANDES
Abstract:	Requirements elicitation is a core activity of requirements engineering for the product to be developed. The knowledge that has been gained during requirements engineering about the product to be developed forms the basis for requirement elicitation. The agile approach is becoming known day by day as the most widely used innovative process in the domain of requirements engineering. Requirements elicitation in agile development faces several challenges. Requirements must be gathered sufficiently to reflect stakeholders' needs. Furthermore, because of the development process, requirements evolve, and they must be adequately treated to keep up with the changing demands of the market and the passage of time. Another challenge with agile implementation is handling non-functional requirements in software development. Addressing non-functional requirements is still a critical factor in the success of any product. Requirements prioritization is also one of the most challenging tasks, and it is uncommon for requirement engineers to be able to specify and document all the requirements at once. This paper presents an approach for requirements elicitation in scrum-based agile development. The approach operates with the feature modeling technique, which is originally used in the Software Product Line (SPL). One of the most important proposed extensions to Feature Models (FMs) is the introduction of feature attributes. Our method uses attributed FMs to consider both functional and non-functional requirements as well as requirement prioritization. For the evaluation purposes, we have demonstrated our approach through two case studies in different domains of software product development. The first case study is in the domain of education, and the second one is in the domain of health care. The results reveal that our approach fits the requirements elicitation process in scrum agile development.
Keywords:	Agile Methodology, Scrum, Attributed Feature Models, Requirements Elicitation, Requirement Engineering, Story Cards.
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	BLOCKCHAIN TECHNOLOGY IN DIGITALIZATION OF RECORDING ACCOUNTING TRANSACTIONS
Author:	MEIRYANI, MARCELINO, TOTO RUSMANTO, THERESIA LESMANA, MOHAMAD IKHSAN MODJO, ADELIA YULMA BUDIARTO
Abstract:	In the industrial era 4.0, the newest digital currency payments emerged, namely cryptocurrency. Cryptocurrency uses blockchain technology to manage and record transactions, so that the financial transaction system does not require a third party as an intermediary and makes transactions more transparent. Blockchain has an impact on the audit process carried out by the auditor. The purpose of this study is to examine the impact of applying blockchain technology in digitizing accounting records. The research method used is descriptive qualitative by finding sources of data from literature, articles, reputable international journals. This research shows that blockchain technology has the potential for implementation in the fields of accounting, auditing and finance. The results of this study indicate that the positive impact of blockchain in digitizing accounting records is the freedom of access to view key records so that auditors can easily authorize transactions, the ease of validating transactions, the guarantee of integrity and reliability of data, the verification process that is carried out automatically , obtain data directly without third parties, ease in providing financial report assessments, have big data analytical features that are guaranteed to be correct, and make the accounting process more extensive and real-time.
Keywords:	Blockchain, Corporate, Accounting, Digitalization, Transactions
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	CLASSIFICATION APPROACH TO PREDICT CUSTOMER DECISION BETWEEN PRODUCT BRANDS BASED ON CUSTOMER PROFILE AND TRANSACTION
Author:	LAURA LAHINDAH, IVAN DIRYANA SUDIRMAN
Abstract:	Businesses need to be able to anticipate what products their customers will buy so that they can better respond to changing market demands and consumer tastes. The purpose of this study is to employ several machine learning models that can reliably estimate the customer's likelihood of purchasing the product given a customer's profile, transaction date, and other transaction information. This was achieved by training and evaluating different machine learning techniques, such as naive bayes, linear models, deep learning, and decision trees, on a dataset consisting of actual transaction data from three months of sales at a medium-scale grocery store in Bandung. Results indicated that naive bayes performed best as a prediction algorithm, this study shows that data mining can be used to predict grocery store datasets. This research provides insights into how machine learning can be used to improve businesses' ability to anticipate consumer behavior and respond to changing market demands. We also found that demographic factors like age and location, as well as contextual factors like time of week, significantly influenced customers' propensity to buy.
Keywords:	Data Mining, Machine Learning, Classification, Naïve Bayes, Customer Product Decision
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	AN OPTIMIZED APPROACH FOR THE DEVELOPMENT OF HIERARCHICAL ENERGY INTEGRATION WITH MULTIPLE ENERGY RESOURCES AND EMPHASIZING THE HEURISTIC TECHNIQUES
Author:	V. SAI GEETHA LAKSHMI, M. DEVIKA RANI, SRAVANTHI KANTAMANENI,PUCHA-NUTHALA SIVAKRISHNA, Dr. D. N. V. SATYANARAYANA, MUTHUKUMAR PARAMASIVAN
Abstract:	Traditionally, electricity has been produced at massive central power plants and distributed to consumers via transmission and distribution systems. However, dispersed generation is quickly becoming the norm. It is called distributed generation when several individual generators are linked together at the distribution level, near the end user. However, this raises new research questions for engineers to consider. Protection and control, self-excitation, and isolation are only a few examples of the difficulties that can arise. In order to investigate these concerns thoroughly, engineers must first configure the foundational simulation model to correctly reflect the distributed generation system. Various renewable energy sources, including wind tu- bines (Type 1 with soft starter capability and Type 3 based on a synchronous generator), solar panels (Pho-tovoltaic Array), and small hydroelectric dams (Synchronous Generator), are evaluated and rated for use in the construction of a microgrid. Here hierarchical methods with heuristic techniques such as harmony search algorithm, simulated annealing algorithm are considered and these algorithms based optimized con-trollers are used for controlling the real and reactive power of various energy sources. This model was cre-ated digitally using PSCAD/EMTDC.
Keywords:	Hierarchical Methods, Multiple Energy Sources, Protection, Heuristic Techniques, Optimized Controllers, Simulation.
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	ANALYTIC APPROACH OF PREDICTING EMPLOYEE ATTRITION USING DATA SCIENCE TECHNIQUES
Author:	R. VINSTON RAJA, A. DEEPAK KUMAR, DR. I. THAMARAI, S. NOOR MOHAMMED, R. RAJESH KANNA
Abstract:	Employee turnover has turn out to be a large venture for data technological know-how companies. The departure of key software program builders would possibly reason large loss an IT business enterprise in view that they additionally leave with essential commercial enterprise understanding and integral technical skills. It is fundamental for IT companies to apprehend developer turnover in order to keep certified builders and reduce injury due to developer exit. In this research, monthly self-report of the software developers includes developer’s activities, working hours, no of projects they have been assigned etc. will been taken into account for analysis for doing the prediction with the help of data science algorithm. By the usage of NB algorithm, KNN algorithm and SVM algorithm, prediction mannequin has been in contrast on the experimental groundwork and supply the end result of which algorithm is performing better. Then, this fantastic mannequin will be given to HR managers to predict whether or not the worker will depart the corporation or not.
Keywords:	Employee Attrition, Naïve Bayes, K-Nearest Neighbor, Support Vector Machine, Turnover Rate
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	STUDY OF E-FILING TAX APPLICATION ACCEPTANCE IN YOGYAKARTA DURING THE COVID-19 PANDEMIC
Author:	DIANA AIRAWATY, WAHYU WIDARJO, RAHMAWATI RAHMAWATI, ARI KUNCARA
Abstract:	Purpose - This exploratory research purpose is to determine tax-payers acceptance of e-filing tax system during Covid-19 pandemic in Indonesia. Design/methodology/approach-The research was conducted on individual taxpayers who live in Yogyakarta for their personal tax reporting using the e-filing tax application. from December 2020 to January 2021 using the google form. The analytical tool used is Smart PLS and using Unified Theory of Acceptance and Use of Technology as a research model. Finding – Performance Expectancy have significant effect on Behavior Intention meanwhile Effort Expectancy and Social Influence have no effect on the intention of using E-filing taxes for individual taxpayers in Yogyakarta. Surprisingly, Facilitating Condition has a positive and significant effect on the intention to use the tax E-filing for personal taxes report and have a positive and significant effect on performance expectancy and effort expectancy, meanwhile social influence has no effect on effort expectancy. Research limitation/implication-This study uses processed data from questionnaires. Differences in respondents' perceptions may occur when answering the questionnaire. Originality - This study develops the UTAUT theory and provide evidence through empirical research to support the finding
Keywords:	Tax E-filing, Personal Income Tax Report, UTAUT
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	ENERGY EFFICIENT HIERARCHICAL AGGLOMERATIVE DATA AGGREGATION IN WIRELESS SENSOR NETWORK
Author:	A.VIMALATHITHAN, Dr.A.SURESH
Abstract:	In wireless sensor networks (WSNs), hierarchical organization structures enjoy the benefit of giving flexible and resource-efficient arrangements. To track down an efficient method for creating clusters, this paper uses a proposed Hierarchical Agglomerative Clustering algorithm by proposing an Energy Efficient Hierarchical Agglomerative Data Aggregation (EEHADA) model. EEHADA model for wireless sensor networks has been proposed. This is an energy-efficient clustering algorithm that utilizes hand-off hubs, variable transmission power, and single message transmission per hub for setting up the group. The problem occurs when sensors transmit data; they use energy in transmission, and uploading the data directly to the sink may require long communication ranges and so degrade the energy of sensors. Indirect communication, in contrast, the data is uploaded to the sink via multiple intermediate sensors, which results in short communication ranges and guarantees the energy efficiency of the sensors. The proposed model is compared with the existing circulated clustering algorithms Drain and HAAS. Simulation results show a significant improvement in the average of the proposed model. The proposed model can be implemented in multi domain in any WSN platform.
Keywords:	Wireless Sensor Network, Energy Efficient Hierarchical Agglomerative Data Aggregation, Clustering Algorithm.
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	MENSEMBLE FEATURE SELECTION AND ENSEMBLE DEEP LEARNING (EDL) CLASSIFIER FOR PARKINSONS DISEASE
Author:	B. SABEENA, S. SIVAKUMARI
Abstract:	PDs (Parkinsons diseases) are chronic neuro-degenerative conditions that impact humans in their day to day lives. Diagnosis and monitoring of these conditions based on limited physical symptoms are painstaking evaluations for medical professionals and clinicians may miss early prodromal phases. Though many DMTs (data mining techniques) for automated assessments of PDs have recently been presented, their performances get reduced due to dataset’s irrelevant features. EFSs (Ensemble Feature Selections) have more benefits than single FSAs (Feature Selection algorithms) as they address drawbacks by mixing different models and improve outcomes of MLTs (machine learning techniques). This work uses OBEFSs (Optimization Based Ensemble Feature Selections) including FMBOAS (Fuzzy Monarch Butterfly Optimization Algorithms), LFCSAs (Levy Flight Cuckoo Search Algorithms), and AFAs (Adaptive Firefly Algorithms) for selection of features based on their correlations. On selection of features, EDL (Ensemble Deep Learning) classifiers classify PD datasets. EDLs include FCBi-LSTMs (Fuzzy Convolution Bi-Directional Long Short-Term Memories), CAEs (Contractive Auto-encoders), and SAEs (Sparse Auto-encoders). CAEs are robust variant of standards of auto encoders which learn representations with reduced sensitiveness to small variations of data. SAEs are used to train classifiers using NNs (neural networks) for identifying PDs from datasets. Stacked generalization is used to combine the results of DL classifiers. When compared to a single model, EDL techniques offer improved predictive performances. The datasets used for this study were obtain ed from machine learning repositories of UCI (University of California-Irvine). The performance of the classifiers were measured using accuracies, F-measures, MCCs (Matthews Correlation Coefficients), and errors.
Keywords:	Parkinsons Disease (PDs), Optimization Based Ensemble Feature Selection (OBEFS), Levy Flight Cuckoo Search Algorithm (LFCSA), and Adaptive Firefly Algorithm (AFA), Ensemble Deep Learning (EDL) Classifier, Fuzzy Convolution Bi-Directional Long Short-Term Memory (FCBi-LSTM), Contractive Autoencoder (CAE), and Sparse Autoencoder (SAE).
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	WALL PATTERN DETECTION WITH PRIM’S ALGORITHM TO CREATE PERFECT RANDOM MAZE
Author:	AGNESIA, WIRAWAN ISTIONO
Abstract:	Replayability is one of the factors that determines how a video game played again by players, with one of the example is by offering new and unique content into a game. One of the methods that can be used is by implementing a map generation into the game level via Procedural Content Generation or PCG. With Prim’s Algorithm as the base of the PCG, this research will design and develop a game with mazes as its map level. Focusing on their maze generation, this research will also on the result and will try to detect and display the data of mazes generated, while also trying to determine video game satisfaction from players that will be playing the game developed via Game User Experience Satisfaction Scale or GUESS. A Detect Wall Pattern method is developed to detect the pattern of each mazes’ grids, where the data will be documented and then processed to determine the result of 250 maze generations of mazes with size 2x2, 3x3 and 4x4. Based on the research, MazeGame has succeeded on being developed with PCG feature based on Prim’s Algorithm. Detect Wall Pattern method has also been developed successfully, where this method successfully detecting 4 unique patterns for 2x2 size mazes, 79 patterns for 3x3 size mazes, and 243 patterns for 4x4 size mazes from 250 maze generation on each size.
Keywords:	Replayability, Procedural Content Generator, Prim’s Algorithm, MazeGame, GUESS, Detect Wall Pattern
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	VISUAL SEMIOTICS ANALYSIS OF OMICRON VARIANT ON INSTAGRAM SOCIAL MEDIA MANAGED BY THE MINISTRY OF HEALTH OF THE REPUBLIC OF INDONESIA
Author:	FARCHAN NOOR RACHMAN, MUHAMMAD ARAS, VIRIENIA PUSPITA
Abstract:	Instagram has been used by the Ministry of Health of the Republic of Indonesia to provide information regarding the Omicron variant of the Covid-19 virus. This is done based on the principle of risk-based communication. It aims to reduce the spread of the Omicron variant of the Covid-19 virus and increase public awareness. The Instagram contents contain visual and verbal messages that focus mainly on spreading information about the new Omicron variant by presenting visual symbols to the public. Through a study using the structuralism method by Ferdinand de Saussure, the visual message is interpreted to give and capture signs in the visual as a representation to get the complete information from the visual message. The structuralism method in analyzing Instagram content of the Ministry of Health of the Republic of Indonesia uses analysis of signifier and signified in each visual and verbal content. The results of this structuralism method show that semiotically, the delivery of messages from the Ministry of Health of the Republic of Indonesia has a key message in the form of increasing public awareness of the spread of the Omicron variant. Furthermore, socializing messages through Instagram content semiotically also includes symbols of Indonesian local wisdom so that the messages given are relevant to Indonesian people.
Keywords:	Social Media, Instagram, Omicron, Structuralism, Semiotics
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	THE EFFECT OF SYSTEM ANALYSIS AND PRODUCT IN DATA PROCESSING ON INFORMATION QUALITY OF FINANCIAL STATEMENTS
Author:	MEIRYANI, RAINA RAHMADANI, HOLLY DEVIARTI, EVI STEELYANA W, FAHRY PRIANDHANA, HUGO PRASETYO, ENGELWATI GANI
Abstract:	Quality accounting information is needed in a company. By producing accounting information that is accountable or open in company information, especially in its financial information to external parties and internal parties. This openness will attract stakeholders' trust in the company. The company's management needs information that can support them in decision making. One of the decisions regarding accounting information. Accounting information systems play a very important role in controlling and securing company assets. This study aims to find out the effect of system analysis and product in data processing (SAP) on quality information of financial statement. The variables tested were system analysis and product in data processing (SAP) on quality information of financial statements. data that collected is secondary data with the method of documentation. The method used in this research is descriptive method. The results in this study system analysis and product in data processing (SAP) affect the relevance of accounting information. The benefits of integrating SAP system modules can save time in every cross-functional business process. This time saving can make companies report their financial statements in a timely manner.
Keywords:	System, Data Processing, Information Quality, Financial Quality, Financial Statements
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	ANALYSIS OF BEST FIRST SEARCH AND FORWARD CHAINING ALGORITHM ON DIAGNOSIS SYSTEM OF BROILER CHICKEN DISEASES
Author:	HANDRIZAL, FAUZAN NURAHMADI, IGO FERDINAND SIHOTANG
Abstract:	Broiler chicken farming is among the most profitable business in Indonesia. Gotland Simatupang Farm is one of the broiler farms in Indonesia. The mortality rate of chickens on this farm is still relatively high. In the last three months, the average mortality proportion was 109.30% whereas in January the total of chicken deaths was 43 chickens and in March the total chicken deaths soared to 90 chickens. After observing the death of chickens, it was caused by several factors. The main cause was disease. Based on the results of discussions with the management of the farm, it was found that there was a lack of veterinarians on the farm. Therefore, a system was built in this study that can help farmers diagnose broiler disease using the Best First Search (BFS) method and the Forward Chaining method. The types of diseases that can be diagnosed in this system only consist of 5 diseases, namely Newcastle Disease, Infectious Bursal Disease, Pullorum, Avian Influenza, and Snot. The system output is the dis disease-affected broiler chickens. System testing is done by comparing the test results with existing sample data. In testing on 20 samples, the actual score % of the accuracy of diagnostic results in the system using the Best First Search method by 70% while using the Forward Chaining method by 90%. Therefore, it can be said that the Forward Chaining method is more accurate than Best First Search.
Keywords:	Best First Search, Forward Chaining, Broiler, Disease
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	ESTIMATION AND MITIGATION OF DIFFERENTAIL MODE NOISE USING EMI FILTERS FOR POWER CONVERTERS
Author:	PATHALA VENKATA SAI CHARISHMA, PAPPU.V. Y JAYASREE
Abstract:	The Electro Magnetic Interference [EMI] problem occurs in electronic devices due to change in switching speed of electronic components. The interference occurs in electronic devices produces noise inside the circuit, the noise decreases the performance of the system and also destroy the components inside the circuit. The noise generated by the equipment has some specific limits as per CISPR STD, so the design of filter has to reduce the noise as per CISPR limits. In this paper Single Ended Primary Inductor Converter [SEPIC] acts as a noise source which is a DC-DC power converter used in many digital applications. The designing of EMI filter for elimination of differential mode noise generated by the power converter by using MULTISIM software and MATLAB. The filter design is made up of passive components called inductors and capacitors. The total setup for estimation of noise from equipment and mitigation of noise with filter consists of Line Impedance Stabilization Network [LISN], Equipment [SEPIC], Noise Separator [Active and Passive noise separators], Power Line filter, PI filter. The LISN is used to produce the constant output impedance in the circuit. The noise generated by the power converter is measured by the noise separators used in the measurement setup. The noise obtained at the output of active and passive noise separator is same with value of 119 dBµV. So, the filter is designed to reduce the noise generated inside the circuit the noise at the output of power line filter is 51.5 dBµV and with PI filter is 52.2 dBµV. The comparison between the filters and calculation of noise from equipment using two type of noise separators are shown in this paper.
Keywords:	Electro Magnetic Interference [EMI], Single Ended Primary Inductor Converter [SEPIC], Power Converters, Line Impedance Stabilization Network [LISN], Noise Separator, Power Line Filter, PI Filter.
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	A STUDY ON AWARENESS AND USAGE OF ICT AND OPEN EDUCATIONAL RESOURCES (OERS) AMONG COLLEGE STUDENTS IN RURAL PARTS OF INDIA
Author:	NAGAIAH M, Dr. THANUSKODI S
Abstract:	This study aims to investigate college students' awareness, availability, and usage of ICT devices and open educational resources (OERs) for academic purposes. The researchers collected data from a stratified random sample of 600 students from rural areas in the Pudukottai district of Tamil Nadu, India, using a questionnaire. The survey found that most students were indirectly using OER in various forms but were not aware of it. The data was analysed using descriptive statistics, and it was found that while students had access to electronic devices and the internet, a lack of awareness about open educational resources was a significant issue. They reported being aware of OERs such as digital learning objects, open-access journals, streaming videos, and learning materials and frequently using them to get relevant learning material, stay up to date on their subject, know the trends in technical fields, watch teaching and learning sessions on online videos, and assist their personal learning. However, they faced challenges such as high internet costs, a lack of technical knowledge, poor network connections, the inadequate availability of hardware and software, and a lack of OERs in their subjects. The report suggests that, while they were aware of some OERs, they were not utilising them to their full potential. The study also found that guidance from staff, social media, and help from friends were the main ways students learned about OERs, and that teachers can greatly help by spreading more information about open educational resources to their students.
Keywords:	Information and Communication Technology, Open Educational Resources, Open Access, Open Education, 5R, Sustainability of OER.
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	A 120 GHZ DOWN CONVERSION MIXER DESIGN FOR IMPROVED LINEARITY, HIGH CONVERSION-GAIN AND LOW NOISE-FIGURE IN 130 NM CMOS TECHNOLOGY
Author:	ZOHAIB HASAN KHAN, SHAILENDRA KUMAR, DEEPAK BALODI, PIYUSH CHARAN
Abstract:	Future mobile networks, like 6G, will use frequencies above 100 GHz. Mobile networks target higher radio frequencies because they provide more bandwidth and capacity. Higher frequencies increase power consumption. This could be a problem for 6G battery-powered phones. High linearity, high conversion gain, and low NF are also challenges. Although Gain, power, linearity, and noise figure are trade-offs. More attention should be paid to the linearity of mixers in the design of wireless receivers because their input signal level is above 100 GHz, which is much higher than any other amplifier in the receiver. This study proposes a Gilbert cell mixer using BALUN to increase linearity, CG, and NF. 120 GHz is the mixer's input frequency. This mixer was simulated using ADS 130 nm RF-CMOS. The RF and LO BALUNS in this Gilbert cell mixer convert 120 GHz to 1 GHz. and obtains 68.664 dB RF-LO isolation, 8.882 CG and 8.4 db NF. The suggested design has -12.2 dBm of 1-dB compression point in this RF bandwidth. At 3 V bias, this double-balanced Gilbert cell mixture dissipates 66.8 mW.
Keywords:	6G, Gilbert Cell Mixer, Conversion Gain (CG), Linearity, 1dB compression point, Noise Figure (NF), BALUN
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	FACTORS AFFECTING INTENTION TO USE OF ONLINE INVESTMENT PLATFORM DURING GLOBAL RECESSION
Author:	NI MADE AULIA SINTA DEWI, TANTY OKTAVIA
Abstract:	A country may face a major downturn in economic activity due to a planned recession in 2023, which could result in more unemployment and lower incomes, as well as affect losses in the investment sector. However, this does not dampen investors' interest in buying and selling investment products online. The purpose of this study is to analyze what factors can influence investors' interest in investing online before a recession occurs. In this study, there are 5 variables studied, including 1 dependent variable (endogenous) from investment interest and 4 independent variables (exogenous) from factors such as investment knowledge, perceived benefits, perceived ease of use, and the last, perception of risk. The researchers then receive up to 170 responses from people who used online investment media using original data derived directly from survey findings sent to the target respondents. For data processing, it uses the SmartPLS program, and the model used is the structural Equation Model (SEM). Based on the results and conclusions of this study, the variables investment knowledge, perceived benefits, and perceived ease of use have a significant positive effect on investment interest, while the perception of risk variable has no significant effect on investment interest in individuals before the recession.
Keywords:	Investment Interest, Investment Knowledge, Perception of Benefits, Perceived Ease of Use, and Risk Perception, Recession
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	QUALITATIVE RECOMMENDER SYSTEM USING ENTROPY-WEIGHTED PEDAGOGICAL CRITERIA FOR EFFECTIVE TRAINING IN E-LEARNING PLATFORMS
Author:	RAJAE ZRIAA1, HICHAM SADIKI, MEROUANE ERTEL, SAID AMALI, NOUR-EDDINE EL FADDOULI
Abstract:	Recommender systems based on collaborative filtering have been widely used in many online learning systems, in order to help learners, find appropriate learning resources. However, these systems are based on classical similarity measures exploiting only learners' ratings for learning objects said subjective preferences, to form groups of learners with similar interests. This paper aims at exploiting also pedagogical criteria of the learning objects, in order to improve the classical similarity measures to generate qualitative recommendations. For this reason, we adopt a Shannon Entropy approach, combining heuristic weights with classical similarity measures, in order to produce recommendations evaluated by their subjective quality, and by their objective usefulness to support learners in their learning process.
Keywords:	Recommendation System; Machine Learning; Hybrid Filtering; Shannon Entropy; Pearson Correlation; E-Learning; Qualitative Recommender System.
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	AN ENSEMBLE MACHINE LEARNING MODEL FOR CLASSIFICATION OF CREDIT CARD FRADULENT TRANSACTIONS
Author:	DR TINA ELIZABETH MATHEW
Abstract:	Digital payment systems such as bank credit cards, debit cards, wallets and more, allow users to make payments anywhere anytime without much hassle and at their convenience. On the other hand, digital payments like Credit card transactions, are vulnerable to many security issues including banking frauds. A credit card user will always prefer a highly reliable system that can detect and prevent banking frauds. Hence techniques that provide better security to these elements or during transactions, identification of genuine and fraudulent transactions are crucial. Machine learning is a promising field of study that can help deal with such critical problems of detection and classification of fraudulent transactions. In this study, the suitability of various machine learning classifiers is investigated in the detection of credit card frauds and, an ensemble machine learning framework which constitutes of a majority voting system implemented on selected classifiers is developed. The performance of the model with feature selection – Pearson Correlation Coefficient and without feature selection is also analyzed. To address the problem of heavy imbalance in the dataset, two class balancing techniques such as Random Under Sampling and Synthetic Minority Oversampling techniques are also applied. The results demonstrate the appropriateness of applying Machine Learning techniques in credit card fraud detection and classification. Digital payment systems such as bank credit cards, debit cards, wallets and more, allow users to make payments anywhere anytime without much hassle and at their convenience. On the other hand, digital payments like Credit card transactions, are vulnerable to many security issues including banking frauds. A credit card user will always prefer a highly reliable system that can detect and prevent banking frauds. Hence techniques that provide better security to these elements or during transactions, identification of genuine and fraudulent transactions are crucial. Machine learning is a promising field of study that can help deal with such critical problems of detection and classification of fraudulent transactions. In this study, the suitability of various machine learning classifiers is investigated in the detection of credit card frauds and, an ensemble machine learning framework which constitutes of a majority voting system implemented on selected classifiers is developed. The performance of the model with feature selection – Pearson Correlation Coefficient and without feature selection is also analyzed. To address the problem of heavy imbalance in the dataset, two class balancing techniques such as Random Under Sampling and Synthetic Minority Oversampling techniques are also applied. The results demonstrate the appropriateness of applying Machine Learning techniques in credit card fraud detection and classification.
Keywords:	Classification, Credit Card Fraud, Ensemble Learning, Machine Learning (ML), Synthetic Minority Over Sampling Technique (SMOTE), Random Under Sampling (RUS).
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	A SECURED BLOCKCHAIN BASED APPROACH FOR DECENTRALIZED AGRI-INSURANCE FOR FOOD CROPS SUPPLY CHAIN
Author:	DAYANA D.S., KALPANA G
Abstract:	Food crop agri-insurance is a critical tool for protecting farmers from the financial risks associated with crop failure or damage caused by natural disasters. By using trustworthy, secured transparent ledgers, smart contracts and decentralized platforms, the efficiency, security, and accessibility of food crop agri-insurance for farmers can be prominently enhanced. By tracking food crop provenance, blockchain aids in the establishment of reliable crop insurance and confidence between farmers and insurance company. Traditional crop insurance schemes are complicated and often unprofitable. Because of the lack of confidence in insurance companies and a concern over deferred and any non-payment of claims, farmers frequently express reluctance to have their crops insured. This paper provides an outline for the potential benefits of using blockchain technology in food crop agri-insurance and proposes the implementation that can automate the claim process, reduce the time and work needed to process claims and make the process faster and more efficient for farmers. It can also increase transparency and reduce the risk of fraud, which improves the overall security of the system. Moreover, the proposed system makes use of smart contracts which are feasible, effective, and inexpensive food crop insurance approach that guarantees farmers are protected and get prompt crop insurance benefits. Also, decentralization and security in blockchain technology ensure transparent transactions between stakeholders and safety in food crop agri-insurance. According to our data, the length of time required for block creation is directly correlated with the time of process.
Keywords:	Blockchain, Crop Insurance, Smart Contract, Distributed Ledger, Latency
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	DEEP LEARNING DETECTING FRAUD IN CREDIT CARD TRANSACTIONS
Author:	IMANE KARKABA, EL MEHDI ADNANI, MOHAMMED ERRITALI
Abstract:	Fraudulent acts -in the financial sector- cause dramatic losses to companies and individuals. To tackle this conundrum, artificial intelligence trends forthcame to develop a fraud detection system. This paper comes to process fraudulent credit card transactions issue deploying ANN and CNN, two supervised deep learning algorithms that proved efficiency .Yet, two hurdles appear: Constant emersion of new fraudulent patterns and highly-imbalanced dataset. So, sampling techniques are required to balance data, the thing that affects the system performance. Thus, Autoencoder, as an unsupervised deep learning algorithm, was added to compare it to the aforementioned algorithms. Three models were trained on a dataset of 284,807 credit card transactions labeled as fraudulent and legitimate. Various techniques were conducted in the pre-processing phase as normalization, data balancing, and feature selection. In the during-applying model stage, tuning and analysis were conducted on the model parameters to improve the classification decision. Similarly, in the post-applying model stage, a boosting technique was applied. Not only were the models compared in terms of accuracy, precision, recall, and AUC score but also they were based on confusion matrix. Eventually, one model was chosen out of the experimented models based on the robustness of detecting new fraud patterns; especially, the latter demonstrates optimal rates, achieving an f1-score of 93% after classifying not fraud transactions.
Keywords:	Unsupervised Deep Learning, ANN, CNN, Autoencoder, Imbalanced Dataset
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	A HYBRID APPROACH FOR ENHANCING THE CLASSIFICATION OF THE PARKINSON’S DISEASE USING SWARM OPTIMIZATION
Author:	MANAL A. ABDEL-FATTAH, RAHMA HUSSEIN EID, AHMED ELSAYED YAKOUB
Abstract:	The second most common neurodegenerative ailment after Alzheimer's disease is Parkinson's disease. It always affects adults from the age of sixty and more. Parkinson's disease symptoms are unnoticed in the early stages, but the symptoms become more apparent as the disease progresses. Recent studies have shown that the disease symptoms can appear in the form of vocal disturbances in the earlier stages and can be used to diagnose Parkinson’s disease. This paper proposes an approach for detecting Parkinson's disease (PD) using speech signals. The motivation of this approach is to improve the detection of Parkinson's early diagnosis by determining the most effective speech examinations instead of produce a huge number of examinations to resistance to the disease at an early stage. As feature selection and swarm algorithms play a vital role during classification, this paper has proposed a hybrid approach based on the Emperor Penguin Colony (EPC) swarm algorithm with Correlation-based Feature Selection (CFC), which is called CEPC. A Parkinson's disease classification dataset consisting of 756 voice measures was used in this study. Before using the proposed approach, five classification algorithms were used to compare accuracy results. Also, the Ensemble classifier has been used in this paper. The CEPC proposed approach provides an improvement in the accuracy of results. An accuracy of 89.4% is obtained by the ensemble classifier, which is higher than some recent work.
Keywords:	Swarm Intelligence Algorithms; Feature Selection; Metaheuristic Algorithms; Emperor Penguins Colony; Classification
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	DATA QUALITY ASSESSMENT USING TDQM FRAMEWORK: A CASE STUDY OF PT AID
Author:	TEDI WAHYUDI, SANI MUHAMMAD ISA
Abstract:	Data has a significant meaning for a business and is used to guide objectives and decision-making. In order to produce high-quality data, a corporation must be able to analyze and manage the data properly. This is a challenge for companies, especially those with a wide variety of data sources, because of the risk of increasing the inaccuracy of the data they have, which can result in making inappropriate decisions. Data processing and acquisition activities are carried out by AID, a company that specializes in "Data as A Service," spanning twelve business units with different business lines that are managed by a conglomerate group that mostly serves the financial services industry. The current state of the organization presents many challenges, as data sources still lack standardization and control, or monitoring of data completeness and accuracy, and the organization has never measured the quality of existing data. In order to obtain quality of data, it is necessary to apply specific methods, processes and techniques, to measure the data. The approach taken in this study to evaluate the quality of the data are Total Data Quality Management (TDQM) and the six dimensions from the DAMA white paper. The results of this evaluation procedure can be used to examine the company's existing data quality and to provide recommendations for changes that need be made internally. The results showed that the quality of data owned by the company was at the threshold of a very high-quality level. Additionally, it is envisaged that this data quality assessment can be applied to all business units and conducted on a regular basis.
Keywords:	Data Quality, Total Data Quality Management, TDQM, data quality dimension, DAMA
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	CNN-MOBILENETV2- DEEP LEARNING-BASED ALZHEIMER’S DISEASE PREDICTION AND CLASSIFICATION
Author:	AMBILI. A.V, DR. A.V. SENTHIL KUMAR, DR. ROHAYA LATIP
Abstract:	Alzheimers disease (AD) is a long-lasting brain disorder for which there is no effective treatment. Yet early detection can delay he growth of the disease. Due to the varied nature of medical tests, manual comparison, visualization and analysis of data can be time-consuming as well as demanding. As a result, an effective method for categorization of Magnetic Resonance Imaging (MRI) images is helpful but extremely difficult. In this paper, the stages of AD are identified using a unique method that effectively classifies brain MRI images using label propagation by involving a Deep Learning (DL)-based framework. Decreased brain tissue volume in brain lobes, hippocampus area, and thalamus are the primary features that aid in differentiating an AD from a normal MRI. The features should be efficient in distinguishing the characteristics between an AD-affected brain and a normal one. A Particle swarm optimization (PSO) based Speed-Up Robust Features (SURF) framework that embeds feature vectors in a subspace to maximize utilization of features that were extracted is presented. A classification method is employed in the newly generated space to categorize data into three classes namely, Normal Condition (NC), MCI, and AD using Convolution Neural Network (CNN)-MobileNetV2. The proposed scheme offers a classification accuracy is 97% yielding a 3% reduced error rate when compared to the best available approaches.
Keywords:	MCI, AD, CNN, MobileNetV2, PSO, SURF
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	ECG BASED MULTI MODAL FRAMEWORK FOR HEART DISEASE DETECTION
Author:	S.IRIN SHERLY, G.MATHIVANAN
Abstract:	Most severe diseases afflicting the human race is as heart disease, so early diagnose and prediction of heart disease is necessary to save human life. Heart disease requires early diagnosis and prognosis in order to save lives. Thus the accurate prediction is done by using deep learning algorithms to avoid the disadvantages in machine learning because which use separate algorithm for feature selection to extract feature. As a result, Convolutional Neural Network (CNN) combined with Aquila Optimization Algorithm (AOA), as Hybrid Deep Convolutional Neural Network (DCNN), which is employed to detect the heart. The AOA algorithm is used to select the weight parameters in the DCNN, which works well on images. Electrocardiogram (ECG) images are used for predicting of cardiovascular disease. The concept driving this study is to combine ECG images as well as clinical data in order to give high performance in prediction. ECG images are pre-processed to scale down the size and then CNN is applied and prediction is done. In this case, different methods for pre processing are applied and best pre-processing method is found in this work. ECG images are taken and different mathematical methods like Fourier transform, DCT or combination of Fourier transform and DCT etc are applied and best method is found. The proposed model is then implemented in MATLAB, and its performance is assessed by comparing it to other existing CNN models.
Keywords:	Deep Convolutional Neural Network, Aquila Optimization Algorithm, Heart Disease, Electrocardiogram, Pre-processing
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	PREDICTING SQL QUERY QUALITY USING MACHINE LEARNING TECHNIQUES
Author:	MOHAMMED RADI
Abstract:	To achieve high database performance (e.g., high throughput and low latency), a database tuning tech-nique is needed to make a database application run faster and respond to end-users on time. Users rely heavily on Structure Query Language (SQL) queries to manage and manipulate their data. The complexity of these queries can range from very simple to very complex. Poorly constructed queries usually lead to performance problems. However, the end-user does not know if the SQL statement is poorly written result-ing in poor database system performance. Therefore evaluating SQL queries can be difficult because there are many syntactic structures for equivalent queries. Manual evaluation is far too time-consuming since there are so many queries in question. Several papers have provided hints and tips on writing good SQL queries to achieve better performance. However, there is a lack of research on identifying poorly-written SQL queries. Therefore, new approaches are needed to automatically identify poorly-written SQL queries, which have to rewritten for faster performance. In this paper, we propose a classification framework to automatically identify well and poorly-written SQL queries. The proposed framework utilizes various ma-chine learning algorithms including Decision Trees, k Nearest Neighbours, Support Vector Machine, and Naive Bayes. In addition, we identified the key features using two different feature extraction techniques namely TFIDF and Count Vectorizer. To effectively evaluate the proposed framework, we used the Delphi technique to manually label two different datasets namely (Bombay and ERPNext). The experimental results demonstrate that the four machine learning classifiers capable to classify the SQL queries into (well, accepted, and poorly) provide promising results in terms of Recall, precision, and F1-score. In both da-tasets, the Decision Trees classifier outperform other classifiers by achieving (90%) on the Bombay Da-taset and (84%) on the ERPNext Dataset in term of F1-measure. Furthermore, the Count Vectorizer out-performs the TFIDF in predicting poorly written queries.Additionally, the proposed framework can serve as a useful tool for database developers and SQL programmers for detecting poorly written query, conse-quently utilized for optimizing SQL query performance.
Keywords:	SQL Query, Machine Learning, Classification, Feature Selection, Database Systems
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	INTEGRATION OF THE MDA APPROACH IN DOCUMENT-ORIENTED NOSQL DATABASES, GENERATION OF A PSM MODEL FROM A PIM MODEL
Author:	AZIZ SRAI, FATIMA GUEROUATE
Abstract:	The volume of data and its diversity are so important today that is the reason why many relational databases are unable to handle this type and large volume of data. To respond to this problem, many NoSQL databases have emerged, such as document-oriented, graph-oriented, key-value-oriented and column-oriented databases. These databases, which revolve around Big Data, have shown an important power in the management of big data. The scientific contribution of the work presented in this article is the application of an MDA approach on a document-oriented approach. We demonstrate the capacity and adaptability of the model-oriented approach on NoSQL approaches, in particular the document-oriented approach. For the presentation of this work we started by introducing the context of our studies which is the MDA approach and the document-oriented approach, then we defined the different metamodels of the sources and targets. We then introduced the different possible model-to-model and model-to-Text transformations using the QVTo transformation language. Finally, we presented as result the document-oriented PSM model and the XMI model generated from the Model to Text transformation with Acceleo. Our motivation for this contribution comes down to the fact that a minority of authors who integrate the concept of programming by model.
Keywords:	Big Data, MDA approach, NoSQL, QVT, PIM model.
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	AN INTELLIGENT SOFT COMPUTING FRAMEWORK FOR FORECASTING STOCK MOVEMENT DIRECTION IN INTERNATIONAL STOCK MARKETS
Author:	K. VENKATESWARARAO AND B. , VENKATA RAMANA REDDY
Abstract:	Every investment in the stock market aims to maximize profit and minimize risk, which is the key to a thriving economy. In today's economy, stock market data prediction and analysis play a crucial role. The non-linear nature of stock market data makes it challenging to analyze. Several studies have been published recently that propose using soft computing techniques to predict stock trades. Predicting consistent outperformance in the stock market has always been a challenge because of the interaction of many factors. In this study, we present an intelligent soft computing framework for forecasting stock movement direction based on efficient feature optimization and hybrid detection technique. First, we introduce an improved Ebola optimization (IEO) algorithm for data preprocessing which filters the noise and irregular patterns from the gathered data. As part of the recursive feature selection, the selection of underlying model hyperparameters is critical. A Chaotic Farmland Fertility (CFF)algorithm is suggested for the feature optimization process which effectively improves the convergence speed for high-dimensional optimization problems. Next, a hybrid Spiking-Quantum Neural Network (hybrid SQNN) framework is present to forecast stock movement direction which ensures avoidance of false prediction rates. Following that, we validate our framework using international benchmark datasetslike the Australian stock market, U.S. stock market and China's wind economy. The simulations have demonstrated the effectiveness of our framework over existing frameworks in terms of error and quality metrics.
Keywords:	Soft Computing, International Stock Markets, Data Preprocessing, Feature Optimization, Stock Movement Forecasting
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	DESIGN ADVENTURE ROLE PLAYING GAME WITH PROCEDURAL CONTENT GENERATION USING PERLIN NOISE ALGORITHM
Author:	RAUL ANDRIAN, ANGGA ADITYA PERMANA, ADHI KUSNADI
Abstract:	Producing quality video game content is not easy. In the process of creating game content, it will require a lot of resources, a long time and expensive costs. This problem can be solved by applying the Procedural Content Generation or PCG method to create game content using certain algorithms without having to create it manually. In this study, the Perlin Noise algorithm was used to create a map for the game that was built. Perlin Noise created realistic visuals and sound effects in games. Besides that, it was very efficient and easy to use. This research was made with the aim of designing and building an Adventure RPG game using the Perlin Noise algorithm, as well as measuring the level of player satisfaction with the game that has been made. The game engine used in the process of making this game is the 2019 version of Unity Engine 3D with the C# programming language on the windows platform. After the game creation process is complete, it will be continued by testing using the Game User Experience Satisfaction Scales 18 (GUESS-18) method which provides 18 questions to measure the level of player satisfaction with games built with PCG. Based on the results of the tests that have been carried out, the final result with a value of 81,85 % indicates that the game that has been successfully built is included in the good predicate category.
Keywords:	Role Playing Game, Procedural Content Generation, Perlin Noise, Game User Experience Satisfaction Scales
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	AN ENHANCED DYNAMIC ALGORITHM FOR ASSOCIATION RULES MINING
Author:	ALA IBRAHIM, MOHAMMAD SHKOUKANI
Abstract:	The data mining techniques have been applied to various fields. Frequent pattern mining is one of the active research themes in data mining. It has an important role in all data mining tasks such as clustering, classification, prediction, and association analysis. A frequent pattern is the most time-consuming process due to the massive number of patterns generated. Frequent patterns are generated by using association rule mining algorithms that use candidate generation and association rules such as the Apriori algorithm. Researchers showed an extraordinary enthusiasm for information mining. Likewise, industrial companies call for different information mining methods to better comprehend client behavior, enhance the service provided and expand business revenues. Due to the huge size of data that exists in databases and warehouses, also because these data are big, dynamic, and change frequently. It is hard and costly to perform mining for frequent patterns and association rules from scratch every time. This paper explores three algorithms. The first one is the classic Apriori algorithm which finds frequent patterns and association rules from scratch every time it works; any update on data will enforce the algorithm to repeat the full process. The second algorithm is the dynamic algorithm which finds frequent item sets by using the accumulated knowledge stored in a database table. This method takes less processing time and computations to find frequent patterns from the data than the classic Apriori. The third algorithm is an enhanced algorithm that we have developed. It includes an indexing technique that helps to discover frequent patterns quickly. The three algorithms were tested on a dataset containing 127936 transactions that contain the items of the market that have been purchased with 207 different types of these market items. The results initially showed that the classical Apriori algorithm was better than the dynamic algorithm and an enhanced algorithm. However, our enhanced algorithm was 78.54% better than the classic Apriori and 27.49% than dynamic algorithm in frequent runs. The results were 169.93, 49.53, and 37.18 seconds for the classic Apriori, the dynamic, and the enhanced algorithms, respectively.
Keywords:	Big data, Data mining, Association rules, Indexing
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text

Title:	DETECTING USERS INTENT BY COMMUNITY QUESTION ANSWERING FOR INFORMATION RETRIEVAL
Author:	MEHDI GAOU, HICHAM TRIBAK, SALAH KRIT, SALMA GAOU
Abstract:	Information retrieval systems (IRs) have seen new types of tools called Community Answering Questions (CQA). It is the taking into account of the need for precise information of the user that motivated the emergence of such systems. A Community Answering Questions (CQA) system can be opposed to an In-ternet search engine like Google or Yahoo! Wiki Answers, Answers and domain-specific forums like Stack Overflow in certain specific points. Although the idea of receiving a direct and targeted response to an issue seems very attractive, the quality of the question itself can have a significant effect on the likelihood of obtaining useful responses. Such an information retrieval paradigm is particularly appealing when the problem cannot be answered directly by the search engines due to the unavailability of relevant online content. A good understanding of the underlying purpose of an issue is essential to meet the information needs of the user better. In this article, we analyze the intent of each question in CQA, the research problem arising from the previ-ously stated objective consists in estimating the best answer according to a question, all its responses and the metadata attached to it. The CQA is reducible to a classification problem, with the "best" answers as a particular class, the rest as a negative class. We can obtain significant and significant improvements in classification concerning state of the art in this field. In addition to textual features, a variety of metadata features are used to model a user's intent, which helps the CQA service better answer to similar questions — recommending the most relevant respondents.
Keywords:	Community Question Answering, Question Retrieval, User Intent, Modeling Entry Point Prdic-tion.
Source:	Journal of Theoretical and Applied Information Technology 15^th May 2023 -- Vol. 101. No. 9-- 2023
Full Text