TIN: Terapan Informatika Nusantara

Penerapan Saliency Maps dalam Explainable AI Untuk Deteksi Penyakit Paru-Paru pada Citra X-Ray Dada dengan Deep Learning

2026-06-10T12:12:09+07:00

Early identification of lung diseases is very important so that medical personnel can quickly provide first aid and further study the patient's condition. In this study, a model was developed to classify chest X-ray images of the lungs using the VGG16 architecture. These chest X-ray images were categorized into three groups: COVID-19, normal lungs, and pneumonia. A combination of hyperparameters, including a learning rate of 0.001, 50 epochs, and a batch size of 16, was used to train the model, achieved an accuracy of 96%. Several evaluation metrics, including precision, recall, f1-score, and confusion matrix, were used to assess the model. In addition, saliency map methods were used to visually interpret the model's prediction output and display the areas of the chest X-ray images that most influenced the model's decision-making. The saliency map visualization findings show that the model focuses its predictions on regions of the lungs associated with the disease, which helps in understanding the algorithm's decision-making process.

Analisis Sentimen Bull dan Bear Market Bitcoin Pada Komentar YouTube Menggunakan Algoritma Support Vector Machine

2026-06-10T12:10:34+07:00

The development of the cryptocurrency market, particularly Bitcoin, significantly influences public opinion expressed through social media platforms such as YouTube. This study aims to analyze bull and bear market sentiments in YouTube comments about Bitcoin using the Support Vector Machine (SVM) algorithm. A total of 25,000 comments were collected using the YouTube Data API. After preprocessing stages including case folding, cleaning, tokenizing, word normalization, and stopword removal, 8,991 valid data were obtained. The dataset was divided into 7,192 training data (80%) and 1,799 testing data (20%). TF-IDF weighting was applied before classification using the SVM algorithm. The evaluation results show that the model achieved an accuracy of 98%, with a macro F1-score of 0.91 and a weighted F1-score of 0.98. Sentiment distribution in the testing data indicates 50.42% neutral (907 comments), 47.08% positive (847 comments), and 2.50% negative (45 comments). The dominance of neutral and positive sentiments reflects relatively stable and optimistic public opinion, consistent with the upward trend of the cryptocurrency market in 2024. This study contributes by providing a YouTube comment-based sentiment analysis approach to describe public opinion tendencies toward Bitcoin bull and bear market conditions, while also offering empirical evidence regarding the effectiveness of the Support Vector Machine algorithm in classifying sentiment within high-dimensional text-based social media data. Furthermore, the findings provide insights into the relationship between public opinion tendencies and cryptocurrency market conditions, which may serve as a reference for understanding market psychology through social media-based sentiment analysis. This study demonstrates that the SVM algorithm is effective in classifying YouTube comment sentiments related to Bitcoin market conditions.

Implementasi Sistem Informasi Penjualan Mebel Berbasis Web Menggunakan Metode SCRUM

2026-06-10T19:29:31+07:00

This study focuses on the design and implementation of a web-based furniture sales application at UD. Andhireyma, Boyolali Regency, Central Java Province. The rationale for developing this software stems from the urgent need to optimize efficiency in daily transaction management, product inventory updates, and sales reporting. Involving the active participation of 14 respondents, including business owners, operational staff, and customers, this study adopted the Scrum framework. The methodology’s phases—product backlog, sprint planning, sprint review, and sprint retrospective—were implemented to ensure that system development iterations could dynamically adapt to user expectations. From a technical perspective, the application architecture was built using the PHP programming language and MySQL database management, integrated with HTML, CSS, and JavaScript elements. Comprehensive data collection was conducted through a series of field observations, continuous documentation, and in-depth interviews. Furthermore, to ensure the overall functionality of all features, the Black Box Testing method was strictly applied. Evaluation findings indicate that the implementation of this digital system successfully accelerated and structured the management of commodity data as well as transaction history reports. The risk of manual recording errors was significantly reduced, which directly corresponds to an improvement in the quality and effectiveness of customer service. Based on the successful functional test results, which recorded a success rate of 80%, this e-commerce platform is recommended and deemed highly suitable for immediate implementation in UD. Andhireyma business operations.

Pengembangan Media Pembelajaran Neuroanatomi Berbasis WebXR (Website Extended Reality) dengan Pendekatan Aksesibilitas dan Optimalisasi

2026-06-10T19:28:10+07:00

This study aims to develop and evaluate a WebXR (Website Extended Reality)-based learning media specifically designed for medical students at Universitas Pendidikan Ganesha. Neuroanatomy learning is a crucial yet challenging component of medical education, often difficult to comprehend due to limited learning media, the scarcity of cadavers, and the abstract nature of the SOBOTA anatomical atlas. The primary contribution of this research lies in the integration of an affordable and easily accessible web-based immersive platform, which simultaneously resolves technical constraints such as motion sickness and stuttering on standalone devices. Behind a well-structured and designed foundation, this media operates on a server-based infrastructure and runs on HTML5 technology architecture to deliver flexible, real-time data accessibility directly through the browser. Technical optimization was implemented in the form of FPS stabilization (FPS Locking) and dynamic resolution scaling to maximize the computing capacity of the Snapdragon XR2 chipset on Meta Quest 2. This application was built using the Multimedia Development Life Cycle (MDLC) framework. The assessment results show that the application has very high validity from both material and media experts, achieving an overall score of 1.00 based on the Gregory matrix. Furthermore, user experience testing using the User Experience Questionnaire (UEQ) with medical students yielded an "Excellent" rating across all dimensions, with the Stimulation dimension receiving the highest score (2.46), placing this application within the top 10% of products globally. The findings indicate that the developed WebXR media effectively offers immersive appeal and can serve as a high-quality, interactive, and cross-platform alternative practical tool for neuroanatomy introduction.

Perbandingan Sensitivitas Serta Stabilitas MQ-4 dan MQ-5 Berbasis Arduino Untuk Deteksi Kebocoran Gas LPG

2026-06-10T19:26:45+07:00

LPG gas leakage is a serious household safety threat in Indonesia, where 88.59% of households use LPG as their primary cooking fuel. Although colorless and faintly odorous at low concentrations, its flammable nature can cause explosions or fires if undetected. This research aims to compare MQ-4 and MQ-5 sensor performance in a single Arduino Uno-based gas leakage detection system. A quantitative experiment with direct comparison was conducted under identical conditions inside a sealed acrylic chamber measuring 16 × 12 × 10 cm, using LPG gas from a refillable lighter to simulate household leakage. Parameters measured include sensitivity, defined as the slope of the log(Rs/R₀) versus log(relative ppm) curve, and reading stability from the standard deviation of repeated measurements, where Rs is sensor resistance when exposed to gas and R₀ is resistance in clean air. All ppm values are relative estimates based on official sensor datasheets. Results show MQ-5 has higher sensitivity with a slope of 0.7914 compared to MQ-4 (0.6505), while MQ-4 is more stable with a log₁₀(Rs/R₀) standard deviation of 0.656 compared to MQ-5 (0.669). The system successfully detected all 30 leakage scenarios (100%) without false alarms, with a classification success rate of 99.0% per session. This research recommends MQ-5 for rapid early detection and MQ-4 for long-term monitoring requiring data consistency, thus both contribute as empirical evidence for a simple, accurate, and economical gas leakage detection solution.

Evaluasi Kualitas Website SEGO PETANI terhadap Kepuasan Pengguna Menggunakan WebQual 4.0

2026-06-10T19:25:34+07:00

This study aims to evaluate the quality of the SEGO PETANI website using the WebQual 4.0 method and analyze its effect on user satisfaction. The SEGO PETANI website is a digital platform designed to help farmers report pest and plant disease attacks in order to support the effectiveness of technology-based agricultural services. The research was conducted at Dinas Pertanian Pangan dan Perikanan Kabupaten Karanganyar involving 167 respondents who had used the SEGO PETANI website. This study employed a quantitative approach with a descriptive-verificative design and purposive sampling technique. Data were collected through questionnaires based on a five-point Likert scale developed from the dimensions of usability quality, information quality, and service interaction quality. Data analysis was performed using SPSS through validity testing, reliability testing, multiple linear regression, t-test, F-test, and coefficient of determination analysis. The results showed that the website quality was categorized as good, with average scores of 4.00 for usability quality, 4.03 for information quality, and 4.06 for service interaction quality. Website quality significantly affected user satisfaction with a coefficient of determination value of 71.1%. Service interaction quality was identified as the most dominant factor influencing user satisfaction. This study provides practical contributions as an evaluation material for improving digital agricultural services and academic contributions through the implementation of the WebQual 4.0 method in public service-based information systems.

Implementasi Retrieval Augmented Generation dan Dynamic Topic Modeling untuk Smart Assistant Berbasis Web

2026-06-10T19:23:45+07:00

Center for research and Community Service (P3M) Politeknik Negeri Manado faces challenges in providing information services that are fast, accurate, and relevant to the academic community. The research title consultation process is still carried out manually and has not been supported by a system that is able to map the latest research trends to provide prospective research topic recommendations. This condition has the potential to cause delays in information and lack of updates on global research developments for lecturers and students. This study aims to develop a web-based Smart Assistant feature that is able to automate P3M information services while providing research title recommendations based on research Trend Analysis. The system was developed by integrating two Artificial Intelligence methods, namely Retrieval Augmented Generation (RAG) to generate chatbot answers based on P3M internal documents and external data through APIs, and Dynamic Topic Modeling using a topical algorithm to analyze research title trends from scientific publication data. The results of the study in the form of Smart Assistant features P3M Manado State Polytechnic, which provides interactive conversation Services and research title recommendations. This system is expected to improve the efficiency of administrative services and help lecturers and students in determining relevant research topics and based on the latest data.

Implementasi Sistem Informasi Manufaktur Berbasis Web dengan Pendekatan Hybrid Rule-Based dan Machine Learning untuk Evaluasi Kinerja Pemasok pada Industri Otomotif

2026-06-11T13:32:46+07:00

Supplier performance evaluation in the automotive manufacturing industry is a critical activity that determines production line continuity. However, this process remains predominantly manual, resulting in administrative inefficiency, data inconsistency, and slow decision-making. This study aims to design and implement a web-based manufacturing information system that integrates a hybrid rule-based and machine learning approach to optimize supplier performance evaluation at PT ABC. The dataset comprises 1,008 transaction records from 28 suppliers over three years (2022–2024) with seven evaluation criteria: Accident, Incident, Line Stop, Off Line, Kanban Delay, Delivery Problem Report (LMD), and Delay Delivery. The research methodology employs Research and Development (R&D) with the Waterfall SDLC model enriched by the CRISP-DM methodology for the analytical component. Feature engineering produced 22 input variables through lag-1, trend analysis, and rolling average techniques, while class imbalance was addressed using SMOTE. Three ensemble algorithms (Random Forest, XGBoost, and Gradient Boosting) were evaluated through 5-Fold Stratified Cross Validation. XGBoost was selected as the best model with 88.82% accuracy and 88.80% Macro F1-Score. The hybrid fusion layer successfully generated tiered action recommendations across five urgency categories, with prediction accuracy on actual operational data reaching 93.16%. The contribution of this research to the development of scientific knowledge is the integration of an AI-based decision support system concept with an operational manufacturing information system platform, while providing a replicable hybrid framework for other manufacturing industry contexts in Indonesia.

Evaluasi Kinerja Algoritma Naïve Bayes, SVM, dan IndoBERT pada Analisis Sentimen Ulasan Pengguna Gojek Berbasis Text Mining

2026-06-11T13:39:38+07:00

This study aims to evaluate and compare the performance of Naïve Bayes, Support Vector Machine (SVM), and IndoBERT algorithms in the task of sentiment classification of user reviews on the Gojek application. Data were collected through web scraping from the Google Play Store and subsequently labeled into three sentiment categories: negative, neutral, and positive. A quantitative approach with a descriptive-comparative design was employed. The research procedure consisted of data collection, text preprocessing, dataset splitting into training and testing sets, model development, and evaluation using accuracy, precision, recall, F1-score, and confusion matrix metrics. The results indicate that the IndoBERT algorithm achieved the best performance, with an accuracy of 92.46%, outperforming Naïve Bayes (87.94%) and SVM (86.43%). Furthermore, IndoBERT demonstrated greater consistency in precision, recall, and F1-score across all sentiment categories. In contrast, Naïve Bayes exhibited a tendency to misclassify certain classes, while SVM showed relatively stable performance, although it did not reach optimal results. These findings suggest that transformer-based approaches are more effective in capturing the contextual complexity of the Indonesian language. This study contributes by providing a comparative analysis of classical and transformer-based methods in Indonesian sentiment classification and offers empirical evidence of the superiority of transformer-based approaches in capturing linguistic contextual nuances in user reviews of digital applications.

Implementasi Sistem Rekomendasi Hybrid untuk Penentuan Reviewer dan Rekomendasi Anggota Tim Peneliti

2026-06-11T13:42:02+07:00

The Center for Research and Community Service (P3M) of Manado State Polytechnic manages a growing volume of research activities. Consequently, a digital transformation in administrative governance is required to ensure temporal efficiency, precise expertise mapping, and compliance with administrative regulations. Beyond objective reviewer assignments, researchers needs a support platform to identify cross-disciplinary collaborators based on publication track records. This study aims to implement a Hybrid Recommender System as an automated solution for reviewer selection and research team formation. The system integrates Content-Based Filtering (TF-IDF and Cosine Similarity algorithms) for text analysis with Constraint-Based Filtering for automated business rule validation. The research methodology follows a Research and Development (R&D) approach using a modified Waterfall model. The processed dataset comprises 516 research titles and profiles of 317 lecturers with SINTA IDs. Black Box testing results confirm the system's effectiveness in validating academic degree qualifications, SINTA score thresholds, and the transparent prevention of conflicts of interest. A comparative evaluation of 44 actual assignment cases demonstrates that the system provides highly relevant recommendations, achieving an accuracy rate of 79.5% compared to manual expert decisions. This research contributes a Microservice-based backend infrastructure that accelerates research coordination while strengthening the transparency and accountability of institutional research management.

Evaluasi Usability pada Website KPU Menggunakan Metode System Usability Scale

2026-06-11T13:43:58+07:00

The digitalization of public services at the General Elections Commission (KPU) is vital for election transparency, yet technical obstacles such as slow loading times and navigation complexity are still frequently reported. This study aims to evaluate the usability of the kpu.go.id website using the System Usability Scale (SUS) method to measure user perception of satisfaction, effectiveness, and efficiency in a standardized manner. This study was conducted in Indonesia with 100 respondents who had previously accessed the kpu.go.id website, using accidental sampling. A quantitative descriptive approach was employed by distributing a questionnaire consisting of ten standard SUS statements. The results show a mean SUS score of 71.5. Based on evaluation parameters, this score places the KPU website in the "Marginal High" category for acceptability, with a "Good" adjective rating and a "B" grade scale. Analysis based on Nielsen's usability dimensions reveals that Satisfaction (81.75) and Memorability (81) perform best, while the Errors (66) and Learnability (66.25) dimensions record the lowest scores. These findings indicate that while the website is functional and acceptable, significant optimization is required in error handling and navigation intuitiveness to bridge the digital divide and ensure service inclusivity for all segments of society. These results serve as a data-driven recommendation for the KPU to enhance its user interface to bolster public trust in democratic information systems.

Analisis Validitas dan Keandalan Tanda Tangan Digital pada Dokumen PDF

2026-06-11T18:06:13+07:00

This research aims to analyze the level of validity, measure consistency as a benchmark, and identify the technical metadata factors that influence the validation results of digital signatures. The research scope includes PDF documents with certificates from official authorities or Certificate Authorities (CA) as well as self-signed ones, tested in a Windows 11 environment using Adobe Reader, Foxit Reader, Google Chrome, and Microsoft Edge. An experimental quantitative method was used through a technical approach with automated data processing based on Python scripts for metadata extraction. The research results show a non-uniform level of validation consistency across applications; Foxit Reader achieved the highest accuracy (96.67%), followed by Adobe Reader (91.67%) and Microsoft Edge (90.00%), while Google Chrome (0%) proved to fail completely due to the absence of a cryptographic computation module. The inconsistencies were identified to stem from differences in parser architectures, local security tolerance limits, and software cipher suite library support. Further diagnostic analysis found vulnerabilities due to the absence of third parties such as a Timestamp Authority, as well as an injection security flaw in Microsoft Edge, which failed to detect directory injection attacks. It can be concluded that the absence of standardised PDF parsing can lead to conflicting cross-platform validation results, the reliability of which is entirely dictated by the integrity of the hash algorithm, the cipher suite, the publisher’s status, and the completeness of the timestamp. The contributions of this research include the development of an interoperability benchmark for PDF digital signature validation, the creation of a consistency matrix for cross-platform evaluation, and the identification of technical metadata factors that are the primary determinants of successful validation. The research findings are expected to serve as a reference for application developers, electronic certificate authorities, and organisations in enhancing the security and consistency of digital document validation.

Perbandingan Performa CNN MobileNetV2 dan K-Nearest Neighbors untuk Klasifikasi Kondisi Tanaman Cabai

2026-06-17T07:30:13+07:00

Diseases and pests affecting chili plants can reduce crop quality and productivity if not detected at an early stage. The identification of plant conditions is still commonly performed through manual visual observation, making the process prone to diagnostic errors and time-consuming. This study aims to compare the performance of a MobileNetV2-based Convolutional Neural Network (CNN) and K-Nearest Neighbors (KNN) for chili plant condition classification using digital images. The dataset consisted of 7,991 training images, 515 validation images, and 355 testing images generated through a cropping process based on bounding-box annotations, covering seven categories of chili plant conditions: Anthracnose, Aphid, Armyworm, Healthy, Leaf Spot, Whitefly, and Yellowisha. Preprocessing included image resizing to 128 × 128 pixels and normalization. The CNN model employed a MobileNetV2 architecture with transfer learning, while the KNN model utilized manually extracted features consisting of Histogram of Oriented Gradients (HOG), Local Binary Pattern (LBP), and Color Histogram. Model performance was evaluated using a confusion matrix, accuracy, precision, recall, and F1-score with a weighted-average approach. The results show that CNN outperformed KNN, achieving an accuracy of 81.69%, precision of 86.28%, recall of 81.69%, and F1-score of 81.34%, whereas KNN achieved an accuracy of 45.63%, precision of 66.80%, recall of 45.63%, and F1-score of 42.24%. The contribution of this study lies in providing a comparative analysis between deep learning and traditional machine learning approaches on a dataset encompassing diseases, pests, and healthy conditions within the same classification scenario. The findings indicate that CNN's automatic feature extraction capability produces more effective visual representations than the manually engineered features used by KNN.

Implementasi Model MaxViT untuk Deteksi Penyakit Daun Cabai Berbasis Mobile

2026-06-17T07:28:55+07:00

Chili is an important horticultural commodity in Indonesia, but its productivity often decreases due to leaf diseases such as Cercospora leaf spot, Powdery Mildew, Mites and Thrips, and Nutritional deficiency. Manual disease identification requires more time and may produce less accurate diagnoses. This study aims to develop a mobile-based chili leaf disease detection system using the Multi-Axis Vision Transformer (MaxViT) model. The dataset consisted of 10,987 chili leaf images divided into five classes and split into training, validation, and testing data with a ratio of 70:15:15. Model training was carried out using four optimizer scenarios, namely a standard baseline model, SGD, Adam, and AdamW. The result showed that the Adam optimizer archieved the best performance with 99,45% accuracy, 99,44% precision, 99,45% recall, and 99,45% F1-Score. The best model was converted into TensorFlow Lite format with a file size of 32.0 MB and deployed in a mobile application. The application can detect diseases through camera capture or gallery images and provide prediction results along with disease descriptions and treatment recommendations. Testing results indicate that the system performs well under various usage conditions. This system is expected to help users indentify chili leaf diseases quickly, practically, and accurately.

Backend Berbasis Laravel dengan Pendekatan Gamifikasi pada Sistem Pelaporan Tugas Harian Mahasiswa

2026-06-17T07:27:43+07:00

Management of digital reporting of students' daily assignments still faces various obstacles, such as data inconsistencies, weak activity validation, delays in collecting assignments, and not yet optimal mechanisms that can improve student discipline on an ongoing basis. These problems show the importance of implementing a backend that is structured and capable of effectively supporting the automation of academic processes. This research aims to implement a web-based daily task reporting system backend using the Laravel framework with a gamification approach in the Nugas system. The research uses a qualitative approach with an implementation case study type and an adaptive Waterfall system development method. The research subjects consisted of 10 active students of the Informatics Study Program at Muhammadiyah Karanganyar University who were selected using a purposive sampling technique. Data collection was carried out through semi-structured interviews, observations, questionnaires and system documentation, while data analysis used source triangulation techniques to maintain the credibility of the research results. Backend implementation produces a system with 17 database Tabels, 15 controllers, and 2 main services, namely GamificationService and NotificationService. The results of black box testing on 74 RESTful API endpoints show that 72 endpoints (97.3%) function according to system requirements. Performance testing using Laravel Debugbar on 10 main endpoints shows an average database query time of ~35 ms without any N+1 query problems found, and the cache implementation runs well. Trials on 10 students showed that all respondents successfully submitted assignments and obtained the Early Bird badge, while the average questionnaire score reached 4.16 on a scale of 5 in the good category. This research provides a contribution in the form of a gamification-based Laravel backend implementation model that is integrated with a RESTful API to support the management of academic tasks in a more structured, automatic and scalable manner. Overall, the research results show that the application of gamification elements in the form of points, badges, leaderboards and achievements can increase student motivation in completing academic assignments.

Rekomendasi Aktivitas Pembelajaran Anak Usia Dini Berbasis Q-Learning dan Profil Perkembangan

2026-06-17T07:26:01+07:00

This study designs a Q-Learning-based recommendation system for early childhood learning activities in a Raudhatul Athfal setting. The system supports teachers in selecting activities aligned with children's developmental profiles, especially physical-motor and cognitive aspects. The recommendation problem is formulated as a Markov Decision Process. The state consists of children's age in months, physical-motor level, cognitive level, previous activity, and previous participation score. The action space contains twelve learning activities, while the reward combines participation, developmental fit, and activity variation. Testing was conducted using scenario-based learning data with five experimental seeds. The results show that Q-Learning achieved an average evaluation reward of 24.667 with a standard deviation of 0.222 from a theoretical scenario bound of 30 points. Ranking evaluation produced Precision@1 of 0.645, Recall@5 of 0.448, and NDCG@5 of 0.641. These results support Q-Learning as a transparent tabular baseline, but they do not prove pedagogical impact. Q-Learning was selected because the state is discrete, the action space is limited, and Q-values are traceable. Since operational data have not been used, the claims are limited to system design, recommendation traceability, and computational evaluation.

Pengembangan Sistem Informasi Dokumentasi Budaya dan Pengelolaan Dokumen Legalitas Kesenian Reyog Ponorogo Berbasis Web Menggunakan Metode Rapid Application Development

2026-06-17T11:02:55+07:00

The management of art studio data and legal documents related to Reyog Ponorogo cultural heritage is still conducted manually, resulting in difficulties in archive retrieval, risks of document loss, and limited public access to cultural information. Furthermore, there is currently no integrated platform capable of combining cultural documentation and legal document management within a centralized digital system. This study aims to develop a web-based information system for cultural documentation and legal document management of Reyog Ponorogo to support cultural digitalization and improve document archiving processes. The research employed the Research and Development (R&D) method using the Rapid Application Development (RAD) model, which consists of requirements analysis, system design, implementation, and testing phases. The system was developed using the Laravel framework, Model View Controller (MVC) architecture, and a MySQL database. The main features include art studio data management, legal document uploads, document verification, cultural news management, and online cultural information services. System evaluation was conducted using Black Box Testing and User Acceptance Testing (UAT). The results of Black Box Testing indicated that all system functions operated according to the specified functional requirements. Furthermore, the UAT results involving 20 respondents achieved a total score of 654 out of a maximum score of 700, resulting in a user acceptance rate of 93.4% and an average score of 4.67 out of 5, categorized as Very Good. These findings demonstrate that the system is capable of supporting cultural documentation and legal document management in a more structured, accessible, and digitally archived manner. The contribution of this study lies in providing an integrated digital platform that combines cultural documentation and legal document management to support the sustainable preservation of Reyog Ponorogo cultural heritage.

Penerapan Algoritma Fuzzy C-Means untuk Pengelompokan Kepuasan Masyarakat terhadap Layanan Berdasarkan Dimensi SERVQUAL

2026-06-17T11:01:18+07:00

Pekanbaru Public Service Mall (MPP) is an integrated service facility that brings together various government agencies in one location. The problem identified is the absence of an in-depth mapping of community satisfaction levels that can realistically represent satisfaction gradations, as the previous approach using K-Means Clustering is crisp in nature and unable to represent the subjective satisfaction of humans who may belong to more than one category simultaneously. Therefore, this study aims to cluster community satisfaction levels toward MPP Pekanbaru services based on five SERVQUAL dimensions using Fuzzy C-Means, and to identify service dimensions that require priority improvement. Unlike K-Means, Fuzzy C-Means allows each respondent to hold membership degrees in multiple clusters simultaneously, making it more suitable for multidimensional satisfaction data. Data were collected through questionnaires distributed to 532 respondents with 23 Likert-scale items (1–5) in accordance with five SERVQUAL dimensions and PermenPANRB Number 14 of 2017. The optimal number of clusters was determined using the Partition Coefficient Index (PCI) by testing four scenarios (c=2, 3, 4, 5). PCI evaluation results showed that c=2 is the optimal configuration with the highest PCI value of 0.799303, achieving convergence at the 12th iteration. Clustering results revealed that 283 respondents (53.2%) belong to Cluster 1 labeled Very Satisfied and 249 respondents (46.8%) belong to Cluster 2 labeled Satisfied. Per-dimension SERVQUAL analysis identified Responsiveness as the primary improvement priority with the largest inter-cluster gap (1.1857 points). The contribution of this research is to produce a Fuzzy C-Means-based community satisfaction clustering model capable of representing satisfaction gradations more realistically than crisp approaches, and to provide a SERVQUAL-based service improvement priority map that can serve as an evaluation reference for MPP Pekanbaru management and other public service institutions.

Rancang Bangun Sistem Informasi Manajemen Service Kendaraan Berbasis Web Bengkel Menggunakan Metode Prototype

2026-06-17T10:59:55+07:00

Vehicle service administration at Syasa Workshop in Semarang is currently suboptimal because the recording of customer profiles, vehicle history, transactions, and spare parts inventory still relies on manual systems. This leads to inefficiencies in data processing, a high risk of document loss, and delays in the preparation of operational reports. This study aims to design and build a web-based vehicle service management information system to accelerate the efficiency and accuracy of workshop data management. The software development adopts the Prototype method, which was chosen because it allows for rapid and adaptive development iterations based on direct feedback from users. The stages of this method include requirements analysis, initial prototype design, periodic user evaluation, iterative improvements, and final implementation. Data collection involved 12 respondents—comprising owners, employees, and customers, through observation, interviews, and document review. System testing using Black Box Testing confirmed that all core functionalities operate according to specifications without technical issues. Meanwhile, the User Acceptance Test (UAT) yielded an acceptability score of 80%, confirming that the system is ready and suitable for implementation. This integrated system successfully digitized the management of customer data, transactions, inventory, booking services, and revenue reporting. Furthermore, the implementation Role-Based Access Control (RPC) not only secures data but also facilitates customers to make reservations independently, while employees can focus on technical management. The main contribution of this research is the provision of a digitalization model for MSME-scale workshop administration that integrates operational management efficiency with ease of customer reservations, thus serving as a practical reference for similar businesses. As a result, workshop administration processes are transformed into faster, more structured, and more efficient.

Rancang Bangun Sistem Informasi Pemesanan Air Galon Berbasis Web Menggunakan Metode RAD

2026-06-17T15:58:24+07:00

The utilization of digital technology has become an essential aspect of improving business management effectiveness, including in the water refill depot sector. At Air Tirta Kencana Depot in Demak Regency, the ordering process and transaction recording are still carried out manually, resulting in slower data processing, an increased risk of recording errors, and customer service that has not been performed optimally. This study aimed to develop a website-based water gallon ordering system by applying the Rapid Application Development (RAD) approach. Research data were collected through observation, interviews, and literature studies. System testing was conducted involving 13 respondents to evaluate system functionality and usability using Black Box Testing and User Acceptance Testing. The developed system was equipped with integrated features for customer data management, transaction ordering, and sales information within a single platform. The integration of these features facilitated ordering, recording, and transaction monitoring processes in a more organized manner according to the operational needs of the depot. In addition, the system enabled users to access order information more quickly and systematically. Based on the testing results, the web-based information system was able to support transaction recording processes to become faster, more organized, and easier to manage. Furthermore, the system achieved a testing success rate of 80% and assisted in monitoring orders and transactions more accurately. The contribution of this research lies in the development of an integrated ordering and transaction management system within a single platform, enabling depot operational processes to be monitored in a more structured manner. Based on these findings, the system also supports more effective business decision-making and helps improve the overall operational quality of the water depot.

Development of a Web-Based Traditional Market Information System Using Waterfall Model

2026-06-17T10:57:17+07:00

The development of information technology drives digital transformation in the trade sector, including traditional markets, which serve as pillars of the people's economy. However, Pasar Anyar market in Tangerang City still faces significant challenges in the efficiency of market data management and in the delivery of commodity price information, which is currently delivered through conventional media. This research aims to design and implement a web-based information system that digitises market operations to improve public access to information. The methodology applied in this development is the Software Development Life Cycle (SDLC) with the Waterfall model, covering systematic stages from needs analysis, system design, code implementation, to testing. The system is built using modern web architecture with CodeIgniter 4 framework, Bootstrap, and MySQL database. The results show that the developed system successfully integrates all market data management, provides real-time commodity price updates, and has an effective news channel related to basic needs. Through functional testing using the Black Box Testing method, this system proved capable of running all main features with a 100% success rate and no errors in the interface logic. This implementation is expected to serve as a foundation for the digital transformation of traditional markets in Tangerang City to improve public service quality.

Analisis Sensitivitas Confidence Threshold pada Semi-Supervised FixMatch untuk Klasifikasi Multi-Kelas Citra Chest X-Ray

2026-06-17T10:55:40+07:00

Optimizing the confidence threshold in pseudo-labeling is a critical technical challenge in Semi-Supervised Learning (SSL) for multi-class medical image classification. A threshold that is too strict limits the utilization of unlabeled data, whereas a threshold that is too lenient introduces low-quality pseudo-labels into the training process. This study applies the FixMatch method with the DenseNet-169 architecture as the backbone network to classify three lung disease categories COVID-19, Pneumonia, and Normal under conditions of extremely limited labeled data. The dataset used is the COVID-19, Pneumonia, and Normal Chest X-Ray Images dataset from Mendeley Data, consisting of 5,218 chest X-ray images, divided into 70% training, 10% validation, and 20% testing sets. The experiments were systematically designed using three labeled-data proportions (5%, 10%, and 15%) and three confidence threshold values (τ = 0.90, 0.95, and 0.99), resulting in nine experimental scenarios. The results demonstrate that τ = 0.95 with 15% labeled data achieved the best performance, obtaining 97.41% accuracy, a 97.49% F1-score, and an AUC of 0.9963. This performance was achieved by balancing pseudo-label selectivity with a sufficient volume of effective training data. At a low labeled-data ratio (5%), the limited amount of labeled data meant that the lower mask rate at τ = 0.95 could not be adequately compensated, allowing τ = 0.99 to perform slightly better. In contrast, at a higher labeled-data ratio (15%), the selectivity of τ = 0.95 produced high-quality pseudo-labels while maintaining sufficient data volume, leading to improved generalization performance. This study contributes an empirical analysis of confidence threshold sensitivity in FixMatch for multi-class chest X-ray classification under limited labeled-data conditions. These findings reveal that the effectiveness of the confidence threshold is highly dependent on the availability of labeled data, and that determining an optimal threshold cannot be separated from the proportion of labeled data available.

Pengembangan Sistem Informasi Manajemen Membership Gym Terintegrasi Kartu Digital Menggunakan Metode Research and Development

2026-06-19T10:17:16+07:00

Data management at Hercules Fitness Karanganyar, which still relies on manual methods using ledgers, serves as the primary background for this research. The problems encountered include the complexity of tracking active membership status, a high potential for duplicated payment records, and the time-consuming preparation of monthly reports, all of which decrease operational productivity. Therefore, this study aims to design and develop a web-based membership information system featuring a novel automated digital card integration and unified notification system utilizing the Laravel framework.This system is expected to facilitate member data administration, transaction recording, attendance tracking, training scheduling, notification delivery, and the provision of digital membership cards. The Research and Development (RnD) approach utilizing the ADDIE (Analysis, Design, Development, Implementation, Evaluation) model was applied in this study, encompassing the phases of needs analysis, system architecture design, development, and evaluation. System testing was conducted at Hercules Fitness Karanganyar, involving 27 participants consisting of gym management and members. Data were collected through direct observation, interviews, and questionnaires. The results indicate that the developed application successfully streamlines the registration process, membership management, transaction tracking, and information delivery to members in a more systematic manner. Based on the Black Box testing, all crucial features on both the administrator and user interfaces operated validly according to the planned scenarios. Furthermore, the User Acceptance Testing (UAT) yielded a feasibility percentage of 88.49%, which based on Likert scale interpretation criteria classifying the system as highly feasible. In conclusion, this Laravel-based membership management software significantly enhances operational efficiency by automating daily administrative workflows and reducing data recapitulation time at Hercules Fitness Karanganyar. This research delivers a valuable practical contribution by providing a digital governance ecosystem blueprint that can be readily adopted by local sports service MSMEs to improve their service efficacy.

Implementasi Speech Synthesis Berbasis Mobile untuk Pelestarian Cerita Rakyat dalam Dialek Melayu Kepulauan Meranti

2026-06-19T10:42:45+07:00

The preservation of Malay folklore faces serious challenges amid the increasing dominance of digital content. This study develops LinguaTales, an Android-based application featuring Malay folklore from the Meranti Islands that integrates Speech Synthesis (Text-to-Speech) technology to provide automated audio narration. The study aims to develop and evaluate LinguaTales while contributing to the preservation of the Meranti Malay dialect, which is increasingly marginalized. The system was developed using the Rapid Application Development (RAD) method with Kotlin in Android Studio. The folklore content was validated by the Malay Customary Institution of Riau (LAMR) Merbau District to ensure the accuracy of dialect vocabulary and cultural values. The application supports three languages (Malay, Indonesian, and English) through two TTS modes: online (Google TTS) and offline (ONNX VITS), supported by a lexical substitution preprocessing mechanism to address dialect vocabulary limitations. Usability testing using the System Usability Scale (SUS) involved 12 respondents, including children, teachers, and parents, resulting in an average score of 77.92 (Acceptable), while the teacher group achieved the highest score of 80.00 (Excellent). The results indicate that LinguaTales is suitable for all target users and contributes to the preservation of the Meranti Malay language and culture through accessible and sustainable digital technology. The proposed lexical substitution mechanism also provides a methodological contribution to the development of TTS systems for low-resource regional languages.

Analisis Komparatif Jarak Euclidean, Manhattan, Canberra, Chebyshev, Cosine pada K-Means untuk Evaluasi Kepuasan Masyarakat

2026-06-19T10:50:40+07:00

The selection of distance metrics in the K-Means Clustering algorithm can affect the quality of clustering results, particularly on public satisfaction data measured using a Likert scale. This study aims to compare the performance of five distance metrics, namely Euclidean Distance, Manhattan Distance, Canberra Distance, Chebyshev Distance, and Cosine Similarity, in clustering the level of public satisfaction toward public services. The research data were obtained from 533 respondents who used the services of the Mal Pelayanan Publik (MPP) Pekanbaru through a questionnaire consisting of 23 questions based on the SERVQUAL dimensions and the Community Satisfaction Survey indicators in accordance with PERMEN PAN-RB Number 14 of 2017. After the data cleaning process, one duplicate record was removed, resulting in 532 respondent records used in the analysis stage. The number of clusters was determined using the Elbow Method, while cluster quality was evaluated using the Davies-Bouldin Index (DBI) and Silhouette Score. The results show that Manhattan Distance with k=2 produced the lowest DBI value of 0.8144, whereas Euclidean Distance with k=3 produced the highest Silhouette Score of 0.5088. The clustering results formed groups of respondents with different satisfaction levels, namely Dissatisfied, Satisfied, and Very Satisfied. This study contributes an evaluative comparison of five distance metrics in the K-Means algorithm using two evaluation approaches simultaneously, namely the Davies-Bouldin Index and Silhouette Score, on public satisfaction data based on a Likert scale. The results indicate that the performance of distance metrics may differ depending on the evaluation method used, therefore the selection of distance metrics should consider the characteristics of the data and the objectives of the analysis.The difference in evaluation results indicates that DBI and Silhouette Score assess clustering quality from different aspects. Based on the findings, Manhattan Distance and Euclidean Distance demonstrated better performance compared to other distance metrics on the dataset used, and can therefore be considered in the analysis of public satisfaction toward public services.

Sistem Informasi Akademik Mobile Berbasis QR Code dan Geolocation untuk Monitoring Kehadiran Siswa

2026-06-22T10:58:16+07:00

Digital transformation in the education sector encourages schools to adopt information systems that improve the efficiency and accuracy of academic data management. This study aims to develop and evaluate a mobile-based academic information system integrated with student attendance validation at SMAN 1 Sliyeg. The main problem identified is the inefficiency of manual attendance recording, which leads to delays in reporting and a high risk of data errors. The system is developed using a client-server architecture that integrates web and mobile platforms. The mobile application is built using Flutter, while the backend uses PHP and MySQL. Attendance validation is implemented through QR Code scanning combined with geolocation to ensure accurate user location verification. The system testing method used in this study is black-box testing to evaluate system functionality and User Acceptance Testing (UAT) to measure user acceptance of the developed system. The test results show that all system features function properly with a success rate of ≥ 90% in functional testing scenarios. In addition, the UAT results indicate that user satisfaction is categorized as good to very good. The results show that the developed system improves the effectiveness of academic data management, reduces recording errors, and accelerates the reporting process. Furthermore, the system enhances transparency and increases parents’ involvement in monitoring students’ academic progress. The main contribution of this study is the development of a mobile-based academic information system that integrates QR Code and geolocation technologies with ±100 meter radius for attendance validation, providing a more accurate, secure, and tamper-resistant attendance mechanism. Furthermore, this study offers a practical model for implementing digital transformation in secondary schools, which can serve as a reference for other educational institutions seeking to develop integrated and mobile-based academic information systems.

The Evaluasi Sistem Pendidikan Digital melalui Kepemimpinan Adaptif: Tinjauan SLR-PRISMA

2026-06-22T11:03:42+07:00

Digital transformation has significantly reshaped various aspects of educational management, including the evaluation of education systems. However, many educational institutions continue to face challenges in optimizing the use of digital technologies due to limitations in human resource competencies, organizational readiness, and leadership capacity to manage change effectively. Therefore, this study aims to analyze the role of adaptive leadership in supporting education system evaluation within the context of digital transformation and to identify the factors influencing its effectiveness. This study employed a Systematic Literature Review (SLR) using the PRISMA framework. Relevant scholarly articles published between 2020 and 2025 were identified through a structured search and selection process. Following the application of inclusion and exclusion criteria, 30 articles were selected for in-depth analysis. Data were analyzed using thematic analysis to identify recurring patterns, trends, and key findings related to adaptive leadership, digital transformation, and educational evaluation. The findings reveal that 24 out of 30 studies emphasized the critical role of adaptive leadership in enhancing organizational flexibility, promoting data-driven decision-making, and fostering technology-based learning innovation. Furthermore, 21 studies reported that digital transformation contributes to more effective educational evaluation through the utilization of information systems and learning analytics. In addition, 18 studies highlighted that the successful implementation of digital transformation is strongly influenced by human resource readiness, organizational culture, and institutional policy support. This study contributes to the literature by providing a comprehensive synthesis of the relationship between adaptive leadership, digital transformation, and education system evaluation, offering insights for policymakers and educational leaders in developing sustainable strategies to improve educational quality in the digital era.

Analisis Komparatif OWASP ZAP dan Nuclei pada Vulnerability Scanning Non-Intrusive Aplikasi Web E-Commerce Publik

2026-06-22T11:09:08+07:00

This study discusses a comparative analysis of non-intrusive vulnerability scanning results on public e-commerce web applications using OWASP ZAP and Nuclei. This study is not intended to directly prove vulnerability exploitation, but rather to evaluate the characteristics of scanning outputs based on the number of aggregate findings, unique findings after deduplication, CVSS v3.1 severity distribution, OWASP Top 10 mapping, overlap, and priority findings that require manual validation. Testing was conducted using a black-box and non-intrusive approach on five targets coded E1 to E5. The coding was applied to maintain testing ethics on public targets, while target selection was based on open web application access, relevance to the e-commerce context, and variations in service characteristics that could be observed externally. The results showed 83 aggregate findings, consisting of 68 OWASP ZAP findings and 15 Nuclei findings. After the normalization and deduplication process, 81 unique findings were obtained with 2 overlapping findings. OWASP ZAP produced more consistent outputs across several targets and was dominant in the Security Misconfiguration category, particularly security headers, Content Security Policy, cache-control, and cookie attributes. Meanwhile, Nuclei produced fewer findings but made an important contribution by detecting 5 Critical findings and 3 High findings, especially on target E4. The limitation of this study lies in output constraints on several targets; therefore, the scanning results cannot be interpreted as the final security condition of the targets, but rather as initial technical indications that require further validation. This study does not measure precision, recall, false positive rate, or scanning time efficiency because the testing was conducted on public targets under non-intrusive limitations and without Proof of Concept. The results indicate that the combination of OWASP ZAP and Nuclei provides more complete analysis coverage than the use of a single scanner because both have different and complementary detection characteristics.

Analisis Pengaruh Sistem Pentanahan Tower dan Sambaran Petir Terhadap Gangguan serta Evaluasi Efektivitas Metode Perbaikan Pertanahan pada SUTT 150 kV

2026-06-22T10:53:55+07:00

The high frequency of lightning activity in Indonesia poses a significant threat to the reliability of power transmission systems. Based on operational data, the 150 kV Tanjung Jati–Sayung transmission line recorded the highest number of lightning-induced disturbances within the Semarang Transmission Maintenance Unit (UPT Semarang) during the 2021–2025 period, making it a relevant subject for further investigation. This study aims to analyze the relationship between lightning strike frequency, tower grounding resistance, and transmission system disturbances, as well as to evaluate the effectiveness of grounding system improvement methods that have been implemented. A quantitative approach was employed using descriptive analysis, Pearson correlation, and simple linear regression based on lightning strike data, tower grounding resistance measurements, and disturbance records. The results indicate that lightning strike frequency has a positive relationship with transmission system disturbances, with a correlation coefficient of 0.137 and a p-value of 0.0452. Similarly, tower grounding resistance exhibits a positive relationship with system disturbances, with a correlation coefficient of 0.2133 and a p-value of 0.0017. Although these relationships are statistically significant, the relatively low coefficient of determination (R² < 5%) suggests that both variables explain only a small portion of the variation in lightning-related disturbances, indicating the presence of other contributing factors that were not included in this study. Nevertheless, grounding system improvement remains one of the technical measures that can be optimized to reduce the risk of transmission line disturbances. The evaluation results show that towers employing the Multi Rod Grounding (MRG) method exhibited the lowest disturbance probability among the grounding improvement methods observed in this study. Overall, the findings provide empirical insights into the relationship between lightning strike frequency, tower grounding resistance, and transmission system disturbances, and may serve as a useful reference for developing maintenance strategies aimed at improving transmission system reliability.