Journal of Informatics and Web Engineering

Editorial Preview

Su-Cheng Haw — 2024-06-13

This editorial highlights all 18 papers in the June issue that deal with the practical aspects of Machine Learning (ML), Artificial Intelligence (AI), Data Mining (DM), the Internet of Things (IoT), Computer Vision, e-learning, and other topics in Computer Science. This issue also includes suggestions for several worthwhile works that deserve further research. With effective from our third volume first issue, we will be publishing triannually in February, June and October.

Facial Skin Analysis in Malaysians using YOLOv5: A Deep Learning Perspective

Ying Huey Gan — 2024-06-13

Nowadays, people are more concerned about their skin conditions and are more willing to spend money and time on facial care routines. The beauty sector market is increasing, and more skin type readers are being created to help people determine their skin type. While various skin type readers are in the market, each is invented and tested abroad. Those skin type readers in the beauty market are not applied well on Malaysian skin. Therefore, this paper proposes a facial skin analysis system tailored primarily for Malaysian skin. This paper integrated object detection and deep learning algorithms in developing skin-type readers. A unique dataset consisting solely of facial images of Malaysian skin was created from scratch for the model. Additionally, You Only Look Once version 5 (YOLOv5) is employed to detect users' facial skin conditions, such as acne, pigment, enlarged pores, uneven skin, blackheads, etc. Then, based on the detected skin conditions, it further classifies the user's skin type into the normal, oily, sensitive, or dry groups.

Classroom Environment Analysis Via Internet of Things

Kai-Yuan Tan — 2024-06-13

In this era of rapid technological advancement, the potential of the digital age has opened up numerous possibilities for our society. However, despite these advancements, traditional classrooms still lack the necessary technology to create an optimal learning environment for students. Consequently, students may struggle to effectively acquire knowledge within classrooms. This paper aims to conduct a classroom environment analysis using Internet of Things technology to gather data and uncover valuable insights. The proposed solution involves an embedded system for controlling and monitoring the classroom environment, as well as exporting historical data for further research. By ensuring accurate data collection, this paper seeks to facilitate meaningful improvements in the classroom environment, aligning with the principle of "garbage in, garbage out" in computer science.

Real Time 3D Internal Building Directory Map

Zi Yang Chia — 2024-06-13

Global Positioning System (GPS) is a famous technology around the world in identifying the real time precise location of any object with the assistance of satellites. The most common application of GPS is the use of outdoor maps. GPS offers efficient, scalable and cost-effective location services. However, this technology is not reliable when the position is in an indoor environment. The signal is very weak or totally lost due to signal attenuation and multipath effects. Among the indoor positioning technologies, WLAN is the most convenient and cost effective. In recent research, machine learning algorithms have become popular and utilized in wireless indoor positioning to achieve better performance. In this paper, different machine learning algorithms are employed to classify different positions in the real-world environment (e.g., Ixora Apartment - House and Multimedia University Malacca – FIST building). Received Signal Strength Indication (RSSI) is collected at each reference point. This data is then used to train the model with hyperparameter tuning. Based on the experiment result, Random Forest achieved 82% accuracy in Ixora Apartment and 84% accuracy in one of the buildings in Multimedia University Malacca. These results outperformed the other models, i.e., K-Nearest Neighbors (KNN) and Support Vector Machine (SVM).

Assessing the Efficiency of Deep Learning Methods for Automated Vehicle Registration Recognition for University Entrance

Muhammad Syaqil Irsyad — 2024-06-13

With the ever-increasing number of vehicles on the road, a faster reliable security system for university entry is needed. This paper presents an approach for Automatic Number Plate Recognition (ANPR) using deep learning and PP-OCRv3. The proposed approach utilizes a pre-trained object detection model to locate license plates, extracts a single frame of the license plate, performs license plate recognition, applies pre-processing techniques, and employs PP-OCRv3 for text extraction in real time. The system was tested with Malaysian vehicle plates, and its accuracy and speed of detection were evaluated. The results show the system's potential to be easily adapted to different camera systems, angles, and lighting conditions by retraining the deep learning model. The paper also explores various deep learning methods, such as CenterNet, EfficientDet, and Faster R-CNN, and their effectiveness in automated vehicle registration detection. The research methodology involves creating a dataset from Open Images Dataset V4, converting label text into XML files, and utilizing the TensorFlow model trained on the COCO dataset. The paper concludes with the synthetic evaluation of the trained models, comparing their performance based on precision, recall, and F1-score. Overall, the proposed approach highlights the potential of deep learning and PP-OCRv3 in achieving accurate and efficient ANPR systems.

Streamlining Dental Clinic Management for Effective Digitisation Productivity and Usability

Sin-Ban Ho — 2024-06-13

Oral health is an integral part of overall health, and poor oral hygiene can lead to a variety of health problems. Modern oral care has greatly improved our quality of life, but the increasing demand for routine dental checkups and treatments calls for improved systems for managing patient records and appointments. While technology has significantly enhanced the efficacy and experience of dental care, many dental clinics still rely on paper records to record the patient’s oral condition, but these are not easily accessible to the patients for viewing. This study aims to address the issue by developing a Dental Clinic Management System to manage patient appointments and records. This system will allow patients to manage their appointments, view their dental history, and receive comments from dentists. Dentists will be able to view appointments, perform treatments, and provide feedback to patients, while the administrator or receptionist will be able to manage appointments, view records, and create invoices. By streamlining dental clinic management, this system aims to improve the overall quality of oral healthcare.

Knowledge-based Word Tokenization System for Urdu

Asif Khan — 2024-06-13

Word tokenization, a foundational step in natural language processing (NLP), is critical for tasks like part-of-speech tagging, named entity recognition, and parsing, as well as various independent NLP applications. In our tech-driven era, the exponential growth of textual data on the World Wide Web demands sophisticated tools for effective processing. Urdu, spoken widely across the globe, is experiencing a surge in, presents unique challenges due to its distinct writing style, the absence of capitalization features, and the prevalence of compound words. This study introduces a novel knowledge-based word tokenization system tailored for Urdu. Central to this system is a maximum matching model with forward and reverse variants, setting it apart from conventional approaches. The novelty of our system lies in its holistic approach, integrating knowledge-based techniques, dual-variant maximum matching, and heightened adaptability to low-resource language speakers, emphasizing the urgent need for advanced Urdu Language Processing (ULP) systems. However, Urdu, labeled as a low-resource language challenges compared to traditional machine learning (ML) approaches. Significantly, our system eliminates the need for a features file and pre-labelled datasets, streamlining the tokenization process. To evaluate the proposed model's efficacy, a comprehensive analysis was conducted on a dataset comprising 100 sentences with 5,000 Urdu words, yielding an impressive accuracy of 97%. This research makes a substantial contribution to Urdu language processing, providing an innovative solution to the complexities posed by the unique linguistic attributes of Urdu tokenization.

Sentiment Analysis in Social Media: A Case Study of Hike in University School Fees in Selected Nigerian Universities

Abdulahi Olarewaju Aremu — 2024-06-13

Faced with escalating operational costs and government disinvestment, Nigerian public universities are implementing tuition fee increases to maintain institutional functionality. This necessary fiscal measure comes in the wake of 2022 industrial action, which exacerbated pre-existing financial strain through extended work stoppages and potentially higher costs associated with resuming activities, while leaving unaddressed the longstanding demands of academics for improved welfare and working conditions. The court-mandated resumption of academic activities without resolution of these core issues further strained university finances, leading to a significant increase in tuition fees. Using VADER, this study investigated social media sentiments related to the increase in university school fees at Usmanu Danfodiyo University, Sokoto, and the University of Maiduguri. The results revealed that students' sentiments regarding the rise in tuition fees at the two universities were largely neutral, with 4.6% positive sentiment, 7.9% negative sentiment, and 87.5% neutral sentiment identified for Usmanu Danfodiyo University, Sokoto. In contrast, the University of Maiduguri had 0% positive sentiment, 19.8% negative sentiment, and 80.2% neutral sentiment. The study recommends seeking feedback through surveys or student leaders and offering scholarships to indigent students to address fee hike concerns at the two universities. While VADER is designed to handle social media textual data, few misclassifications of sentiments were noted and discussed.

A Bilingual Immersive Environment for Kids Learning

Tufail Ahmad — 2024-06-13

Virtual Worlds (VWs) provide an improved learning experience as they have an increased level of user participation and technological involvement. They avoid content duplication and following a set of predefined rules to overcome the issues in game environments. VWs are coherent, persistent and collaborative social places that believe in realism through immersion. Client viewer software is used for user immersion in the form of avatars and in-world content creation in these environments. This work presents a simple bi-lingual environment developed for kids’ learning in English and Urdu using the well-known VW development framework called OpenSimulator (OpSim). It used Firestorm viewer for in-world content development and Linden Scripting Language for making the content dynamic and interactive. This work used Blender, Adobe Photoshop and Illustrator for developing images of different objects and adding their related dynamics before integrating them within the OpSim. It developed some basic activities for learning about colours, national personalities of Pakistan, Urdu alphabets, geometrical shapes and fruits. Bots were used to populate the content for making it more appealing. The proposed environment provides an arbitrary number of tries to perform each activity and guides kids through a real time positive feedback towards an improved learning. This work conducted initial validation tests of the proposed presence on standalone mode of OpSim framework with the help of domain experts, which confirmed the effectiveness of the proposed mechanism. However, it was suggested to conduct further validation on grid mode of OpSim framework. The developed environment shall be compared with other methods for learning purposes. It could incorporate more languages, activities and lessons. 3D working models of the alphabets, fruits and vegetables, and customized avatars capable of interacting with kids would offer more positive impact on learning. Similarly, the simple tasks could be extended to multiplayer collaborative games.

Unveiling the Efficacy of AI-based Algorithms in Phishing Attack Detection

Tajamul Shahzad — 2024-06-13

Phishing poses a significant challenge in an ever-evolving world. The increased usage of the Internet has resulted in the emergence of a different kind of theft referred to as cybercrime. The term cybercrime describes the act of invading privacy and illegitimately obtaining personal information using digital platform. Primarily an approach named phishing is employed, which involves the use of spoof emails or bogus websites by the attackers to get the victim's personal information like their account credentials, debit, or credit card’s number, etc. To give the brief knowledge of phishing attacks and their types of the objective of this work is to investigate various AI algorithms. Through a detail literature 14 AI algorithms which are repeatedly used for detection, and these are Random Forests, Convolutional Neural Network, Naïve Bayes, K-Nearest Neighbours algorithm, Decision Trees, long short-term memory, gated recurrent unit, Artificial Neural Network, AdaBoost, Logistic Regression, Gradient Boost, Multi-layer perceptron, Recurrent Neural Network, Extreme gradient boosting, and Support Vector Machine to detect phishing attacks. To verify the effectiveness of these algorithms an experiment is performed on two datasets. Among all the algorithms Convolutional Neural Network, Multi-layer perceptron and AdaBoost achieved more than 90% accuracy, precision and sensitivity and it was showed through results that these algorithms are very efficient and can achieve high accuracy if used to the requirements of specific scenario with proper planning. Moreover, the paper shows how different AI techniques have been employed in multiple studies to detect and address phishing attacks. Also, this paper gives a complete list of current problems with phishing attacks and ideas for future studies in this area.

Ensemble Learning-Powered URL Phishing Detection: A Performance Driven Approach

Shougfta Mushtaq — 2024-06-13

With the rapid growth in the usage of the Internet, criminals have found new ways to engage in cyber-attacks. The most common and widespread attack is URL phishing. The proposed system focuses on improving phishing website detection using feature selection and ensemble learning. This model uses two datasets, DS-30 and DS-50, each with 30 and 50 features. Ensemble learning using a voting classifier was then applied to train the model, achieving more accuracy. The combination of HEFS with random forest distribution achieved 94.6% accuracy while minimizing the number of features used (20.8% of the base feature set). The classifier works in the proposed model, and the accuracy is 96% and 98% on the DS-30 and DS-50 datasets, respectively. The hybrid model uses a combination of different factors to distinguish phishing websites from legitimate websites.

Secure Room-Sharing Decentralized App Development on Ethereum Block Chain Using Smart Contracts

Hasnain Raza — 2024-06-13

The purpose of this research is to analyze whether Blockchain technology can affect the share-economy. Apart from that, blockchain technology has been innovating the whole of the industries and so the academics are discovering the possibilities and starting to incorporate them in order to provide additional tech possibilities. The sharing economic system is the system which enables to share asset among the one person to the other person. It has seen the remarkable growth in the last few years, Uber, Careem, Airbnb, Zostel, Hostel World are some companies to mention which have fueled this growth. Yet, the majority of the transactions through the sharing economy system are facilitated by a centralized infrastructure executing an intermediary role that might be vulnerable to issues of hacking and data breach and such operations come at a high cost and expending more effort in keeping the system active is also a factor worth mentioning. A different method which is free of control centers such as the peer-to-peer sharing and smart service model which is being implemented in the Hospitality industry can overcome those obstacles. Through the use of a blockchain-backed payment system based on an accommodation-sharing structure, the research will develop a prototype of the proposed system in the form of a DApp on the Ethereum blockchain. The aim of these studies and research is to inform the public about the revolution that is blockchain and its benefits for trade, technology, business, and daily life.

Temporal Climatic Shifts in Henan Province: A 16-decades Perspective Through Regression, SARIMA, and NAR Modeling

Lin Qing — 2024-06-13

Global warming is having a significant impact on all aspects of human production and life. This study employs a cross-sectional analysis to investigate the temporal dynamics of average temperature changes in Henan Province, China, from 1851 to 2012. Utilizing the Berkeley Earth Surface Temperature Data and the Daily Meteorological Dataset of China National Surface Weather Station v3.0, we applied regression analysis, Seasonal Autoregressive Integrated Moving Average (SARIMA), and Nonlinear Autoregressive Network (NAR) models to predict temperature trends. Results indicate a significant warming trend over the 160-year period, with the models demonstrating strong predictive performance, albeit with some variability. The study underscores the increasing temperatures' implications for the province's agricultural sustainability and ecological balance. This study highlights the urgency of understanding and mitigating climate change's impacts, particularly in Henan Province, China, for the sake of agricultural sustainability, water resources, and public health. The research findings contribute valuable insights and methodologies to climate data analysis, aiding future predictions and policy-making efforts.

Social Messaging Application with Translation and Speech-to-Text Transformation

Kang Qin Yip — 2024-06-13

Unlike traditional SMS or MMS, messaging apps offer a broader range of data transmission capabilities. The application utilizes a WIFI or internet connection and enables users to exchange information through various means such as text, voice, and multimedia files. However, popular messaging applications such as WeChat, Telegram, and WhatsApp have limitations in language translation and file uploading size. Thus, this project aims to address these limitations by developing a social messaging application that serves as a comprehensive communication tool. The application will facilitate both written and verbal communication by providing translation services for various languages, including voice messages. The proposed application intends to act as a versatile platform, functioning as a translator while enabling seamless communication between users in different languages. Translation accuracy and BLEU metric are applied to evaluate the efficacy of the enhanced social messaging application. The proposed application is able to translate voice and written messages into another language with the help of Google translation API as well as Speech to text API. The BLEU average score between English and Malay is 0.94 but the translation between Malay and Chinese is 0.82, Chinese and English is 0.70. Though not perfect, the proposed application can enhance the current social messaging application with the speech-to-text feature and message translation feature. Last but not least, a concluding remark is provided to further improve the application in future.

Development and Validation of Autotronic Training Module for Automobile Technology Students in Polytechnics in Southern Nigeria

Saue, Baritule Prince — 2024-06-13

A research was carried out to create and verify an autotronic training module for students studying vehicle technology at polytechnics located in Southern Nigeria. The design used for the project was Research and Development (R&D). An observation has been made that the curriculum of polytechnics in Nigeria lacks sufficient substance on autotronics technology. As a result, lecturers have challenges in fully imparting the abilities that are essential for the professional world. Undoubtedly, there is now a disparity between the training that craftsmen receive and the skills that are demanded by industries. The research was carried out in the southern region of Nigeria. The research focused on a specific demographic of 1,443 respondents. All the lecturers teaching automobile technology were included in the study, and a purposive sampling technique was used to select 75 automobile technicians from the region. A grand total of 122 questionnaires were distributed to the respondents with the assistance of five research assistants, one hailing from each state. The researcher created an 86-item questionnaire called the Autotronic Training Module Questionnaire (ATMQ). The 5-point Likert scale includes answer alternatives such as Highly Appropriate (HA) - 5, Appropriate (A) - 4, Moderately Appropriate (MA) - 3, Inappropriate (I) - 2, and Highly Inappropriate (HI) - 1, respectively. The tools (Questionnaire and Multiple Choice Questions) underwent face and content validity assessment by three (3) experts. The instrument's internal consistency was assessed using Cronbach Alpha reliability, resulting in a coefficient of .88. The collected data were analysed using the mean and standard deviation to address the study objectives. Additionally, the hypotheses were tested using Analysis of Covariance (ANCOVA). A criteria mean of 3.5 or above was considered 'acceptable', while anything below was considered 'inappropriate'. In addition, the F-calculated (F-cal) ratio was compared to the .05 probability level of significance for each hypothesis. If the F-ratio is lower than the .05 probability level of significance, the null hypothesis was rejected; otherwise, it was accepted. The research concluded that the goals, materials, training facilities, training method, instructors' activities, students' activities, and assessment procedures are suitable for incorporation into the autotronic training module in polytechnics in Southern Nigeria.

Performance Evaluation of YOLO Models in Plant Disease Detection

Usman Ali — 2024-06-13

Plant diseases significantly impact global agriculture, leading to substantial production losses and economic consequences. Timely disease detection can enhance crop yield, optimize resource utilization, reduce costs, and mitigate environmental effects, ultimately ensuring high-quality food production. Deep learning, specifically computer vision-based techniques, have proven invaluable in tasks like image classification, segmentation, and object detection. Deep Learning techniques such as You Only Look Once (YOLO) models are state of the art neural network algorithms used for accurate object detection. In this study, YOLOv5, YOLOv7 and YOLOv8 models were trained on CCL’20 dataset for citrus disease detection. Data augmentation techniques such as image translation, image scaling, flip, mosaic augmentations were implemented to improve the models’ performance during training phase. The model performance was evaluated using metric such as Mean Average Precision at 50% to 95% Intersection over Union score i.e. mAP@50-95. The results show that YOLOv8 model performs better than other variants and offers significant improvements over the benchmark performance from previous studies. The final hyper-parameter tuned model achieved 96.1% mAP@50-95 on testing data for citrus disease detection and mAP@50-95 of 95.3%, 96.0% and 97.0% for detection of Anthracnose, Melanose and Bacterial Brown Spot diseases, respectively. The trained model was able to detect single and multiple instances of same or different disease in an image showing the potential of recent YOLO models. The trained YOLOv8 model is deployed on Roboflow platform.

Wix for Web Development and the Application of the Waterfall Model and Project Based Learning for Project Completion: A Case Study

Mawar Madiah — 2024-06-13

Website development without prior knowledge of HTML or programming experience would be a significant challenge. This study aims to share the students' experiences developing a website using Wix, a user-friendly website builder. The website was assigned as a project-based assessment of one of the courses required to complete a Foundation programme. The course was delivered using a project-based learning (PBL) approach in this context. The students worked as a group to write a project proposal, plan activities, develop the website, write reports, and present the outcome. As for the website development process, the study demonstrates the completion of the project using the Waterfall model, a sequential approach to the software development lifecycle (SDLC). In summary, this study aims to explore the use of Wix for web development, the PBL approach for course delivery, and the Waterfall model for completing the website project. The result shows that the students could complete the website development project in a timely manner by using Wix as their web design platform and adopting the Waterfall model as their project management approach. In addition, the students also benefitted from the PBL approach. The outcomes of this study will be of great benefit to educators, particularly in their role in helping students complete their website development projects.

Ensemble-SMOTE: Mitigating Class Imbalance in Graduate on Time Detection

Theng-Jia Law — 2024-06-13

In education, detecting students graduating on time is difficult due to high data complexity. Researchers have employed various approaches in identifying on-time graduation with Machine Learning, but it remains a challenging task due to the class imbalance in the dataset. This study has aimed to (i) compare various class imbalance treatment methods with different sampling ratios, (ii) propose an ensemble class imbalance treatment method in mitigating the problem of class imbalance, and (iii) develop and evaluate predictive models in identifying the likelihood of students graduating on time during their studies in university. The dataset is collected from 4007 graduates of a university from year 2021 and 2022 with 41 variables. After feature selection, various class imbalance treatment methods were compared with different sampling ratios ranging from 50% to 90%. Moreover, Ensemble-SMOTE is proposed to aggregate the dataset generated by Synthetic Minority Oversampling Technique variants in mitigating the problem of class imbalance effectively. The dataset generated by class imbalance treatment methods were used as the input of the predictive models in detecting on-time graduation. The predictive models were evaluated based on accuracy, precision, recall, F0.5-score, F1-score, F2-score, Area under the Curve, and Area Under the Precision-Recall Curve. Based on the findings, Logistic Regression with Ensemble-SMOTE outperformed other predictive models, and class imbalance treatment methods by achieving the highest average accuracy (87.24), recall (92.50%), F1-score (91.30%), and F2-score (92.02%) from 6^th until 10^th trimester. To assess the effectiveness of class imbalance treatment methods, Friedman test is performed to determine on significant difference between the models after applying Shapiro-Wilk test in normality test. Consequently, Ensemble-SMOTE is ranked as the top-performers by achieving the lowest value in the average rank based on the performance metrics. Additional research could incorporate and examine more complicated approaches in mitigating class imbalance when the dataset is highly imbalanced.

Empirical Analysis of CI/CD Tools Usage in GitHub Actions Workflows

Adam Rafif Faqih — 2024-06-13

As software systems grow larger and more complex, with rapidly changing requirements, manually managing code integration, testing, and deployment becomes extremely challenging. Continuous Integration and Continuous Deployment (CI/CD) practices and tools have emerged to help automate these processes. This research explores the usage of different categories of CI/CD tools within GitHub Actions workflow configurations across GitHub repositories. The five-tool categories analyzed are Version Control Management, Static Code Analysis, Build Automation, Test Automation, and CI/CD Servers. The data used in this research is from a dataset of GitHub Actions workflow configuration files. From the data, the usage is extracted and the concurrent usage of the tools is calculated. Next, the tools are labeled based on their taxonomy. In our finding, the build automation has the biggest number of uses, while the test automation has the least number of uses. Our finding indicates the correlation between the tool category and the programming language used in the software project. Meanwhile, some tools cannot be classified into the existing taxonomy. This can lead to reevaluating the taxonomy structure of CI/CD tools.