Proceedings of the National Conference on Electrical Engineering, Informatics, Industrial Technology, and Creative Media

Program Book

Andi Prademon Yunus — Thu, 22 Jan 2026 00:00:00 +0800

Advance solution for local and global problems through multidisciplinary and industrial synergy

Usability-Driven E-Commerce for EcoFashion: The Cimemo.id Redesign Case

Sukhaenah Tri Utami, Tenia Wahyuningrum, Khoem Sambath — Wed, 28 Jan 2026 11:46:57 +0800

Eco-friendly products are a solution to the widespread textile industry waste pollution in Indonesia. To support the eco-fashion trend, Cimemo.id, an ecoprint boutique established in 2018 in Purwokerto, faces challenges in attracting buyers and introducing ecoprint products more widely. Furthermore, Cimemo.id requires a platform that can address these challenges. The existing mobile application has drawbacks, including a lack of flexibility due to the need to install it on the user's mobile phone. Therefore, alternatives are needed to improve the user experience. This study aims to enhance the user interface and experience by redesigning the Cimemo.id e-commerce website using the User-Centered Design method. The design process involved five UCD stages and usability testing with 70 respondents, determined using the Slovin formula. Evaluation was conducted using the System Usability Scale to measure effectiveness, efficiency, and user satisfaction. The results showed an effectiveness rate of 98.36% (very effective), an efficiency of 0.184 goals/second (very fast), and a SUS score of 87.64 (A-Very Good). Its effectiveness exceeded user performance testing in previous research by 15%, and its efficiency increased by 11%. The front-end website implementation was tested using black box testing, with a 'pass' result for all components. This research yielded an effective interface design, ready for further development with the addition of a back-end system to achieve full functionality.

ClimatePulse: Sentiment and Emotion Analysis of Public Discourse on Climate Change in Social Media using BERT, NER, Multilabel Classification, and Spatio-Temporal Visualization

Wed, 28 Jan 2026 12:07:19 +0800

Climate change poses major global challenges with wide-ranging impacts on ecosystems, health, and policy. Public responses to climate policies are increasingly voiced on social media, producing large volumes of data that require intelligent analysis. This study introduces ClimatePulse, a unified system for analyzing Indonesian public opinion on climate change by integrating sentiment analysis, fine-grained emotion classification, and spatio-temporal visualization. The system leverages a fine-tuned BERT model trained on a balanced dataset of 3,256 tweets. Results demonstrate that the model achieves an accuracy of 75.57% and a Macro F1-Score of 0.7556, outperforming traditional baselines like SVM and Logistic Regression. Specifically, the model excels in detecting negative sentiment with an F1-Score of 0.8151, capturing critical public dissatisfaction. Beyond sentiment, the system identifies dominant emotions (e.g., sadness, joy, fear) and visualizes geographic trends through an interactive map, providing actionable insights for policymakers. While challenges remain in classifying ambiguous neutral texts (F1-Score 0.7035) and detecting sarcasm, ClimatePulse effectively bridges the gap between unstructured social media data and data-driven decision-making, directly supporting SDG 13: Climate Action.

Whistleblowing System for Website-Based Sexual Violence Complaints: An Improved Approach Combining Design Thinking and Software Development Life Cycle

Luthfiah Gustiana, Tenia Wahyuningrum, Teotino Gomes Soares — Wed, 28 Jan 2026 11:57:35 +0800

Cases of sexual violence in higher education often go unreported due to concerns about reporter identity safety and the absence of adequate reporting mechanisms. This study developed a website-based sexual violence reporting system equipped with whistleblowing features, using a combined approach of Design Thinking and the Software Development Life Cycle (SDLC). The Design Thinking stages were applied to understand user needs and design empathy-based solutions, while SDLC provided the technical framework for system implementation. Development followed an iterative process, starting with planning, UI/UX design, and interface development using React.js, and then proceeding to system testing through black-box testing, usability testing, and the System Usability Scale (SUS). The testing results show that the system functions as intended and achieved a SUS score of 82.43 (Excellent category) from the user side and 79.5 (Good category) from the admin side, indicating that the system is easy to use and fosters a sense of safety and comfort for reporters. Overall, the system shows potential to enhance comfort, security, and the effectiveness of digital reporting for sexual violence cases.

Determining Factors in the Success of the SatuSehat Application to Support the Free Health Program Using the DeLone & McLean Method

Desi Rahmawati, Adnan Purwanto, Muhammad Akbar Setiawan, Riana Safitri — Wed, 28 Jan 2026 12:33:53 +0800

SatuSehat Mobile is a digital health platform introduced by Indonesia’s Ministry of Health as a key initiative in the country’s ongoing healthcare technology modernization. This application provides various features to help users manage and monitor their personal health, one of which is the Free Health Check feature that functions for early detection of non-communicable diseases such as high cholesterol, diabetes, hypertension, lung function disorders, and early cancer screening. This study aims to evaluate the impact of the Free Health Check feature on users and measure the effectiveness of the SatuSehat Mobile application in achieving its objectives using the DeLone & McLean method. This method includes six measurement variables, namely: system quality of SatuSehat Mobile, information quality of SatuSehat Mobile, service quality of SatuSehat Mobile, use of SatuSehat Mobile, user satisfaction of SatuSehat Mobile, and net benefits of SatuSehatMobile. Data analysis was performed using the Smart PLS-SEM technique to test the hypothesis. The results of the PLS-SEM analysis show that of 9 relationships between constructs that were tested, there were 5 significant relationships, namely Information Quality and Service Quality to Use, System Quality to User Satisfaction, Use to Net Benefits, User Satisfaction to Net Benefits. There were 4 insignificant relationships, namely Information Quality, Service Quality and Use to User Satisfaction, System Quality to Use.

Optimization of Random Forest Model with Correlation-Based Feature Selection for Enhanced Forest Health Prediction

Wed, 28 Jan 2026 12:42:16 +0800

Forest health serves as a key indicator for maintaining ecosystem sustainability and biodiversity. This study aims to predict forest health status using a Random Forest algorithm integrated with Correlation-Based Feature Selection (CFS). The dataset comprises 1,000 samples with 18 attributes—including Disturbance_Level, Fire_Risk_Index, Tree_Height, and Menhinick_Index—along with health status labels categorized into four classes: Unhealthy, Sub-Healthy, Healthy, and Very Healthy. The research methodology encompassed data preprocessing, feature selection using CFS, Random Forest model construction, and performance evaluation. Feature selection identified four key attributes that significantly contributed to forest health prediction. The model was trained on 70% of the data and tested on the remaining 30%, achieving an accuracy of 92%. Further analysis revealed an average precision of 91%, recall of 90%, and F1-score of 90%. The confusion matrix indicated accurate predictions across most categories, though some misclassification occurred in the Sub-Healthy class. This study demonstrates that the CFS-based Random Forest approach is effective for forest health prediction, offering a valuable analytical tool to support conservation efforts and damage risk mitigation.

, , , Predictive Modeling

Clustering Passenger Satisfaction Levels in Air Travel Using the K-Means Clustering Algorithm

Pipin Tri Hastuti, Dwi Hartanti — Wed, 28 Jan 2026 13:21:51 +0800

This study aims to cluster the satisfaction levels of airline passengers in the business class segment with business travel purposes who are categorized as disloyal, using the K-Means clustering method. The data was sourced from the Airline Passenger Satisfaction dataset on Kaggle, then cleaned, filtered for disloyal business travelers, and transformed into numerical format. The optimal number of clusters was determined using the Elbow Method, which indicated an optimal value at k=3. Clustering was subsequently carried out with the K-Means algorithm and visualized using PCA. Cluster quality evaluation employed the Davies-Bouldin Index, resulting in a value of -0.5, indicating reasonably good cluster separation. These findings can help airlines understand patterns of dissatisfaction among premium customers and design more targeted service strategies to improve their loyalty.

^1*, Dwi Hartanti

Designing and Validating a Website-Based Knowledge Management System for Micro, Small, and Medium Enterprises: A KMSLC Approach with SECI-Driven Knowledge Capture

Johanes Dom Noel Wijaya, Muhammad Nurwegiono, Rudy Setiawan — Wed, 28 Jan 2026 13:36:36 +0800

This study develops a website-based Knowledge Management System (KMS) for an MSME (Amadea Kitchen) by combining the Knowledge Management System Life Cycle (KMSLC) focused on knowledge capture, design blueprint, and verification & validation with SECI-driven mechanisms to surface and codify tacit and explicit knowledge. The captured assets were translated into a minimal, deployable blueprint comprising a knowledge map, role-based actors (admin/owner/employee), and three high-leverage modules (Document Library, FAQ, Feedback). Verification confirmed that explicit artifacts were digitized (PDF) and recurring tacit themes were codified into FAQs, while validation addressed usability and operational performance. With eight participants (one owner, seven employees), the system achieved a SUS score of 79.6 ± 6.8 (95% CI [73.9, 85.3]), indicating Good usability. Performance measurements over 10 runs per page yielded average page loads of 2.10 ± 0.22 s (Homepage), 2.32 ± 0.27 s (Document Library), and 1.84 ± 0.19 s (FAQ), with server response times of 186 ± 21 ms, 204 ± 25 ms, and 175 ± 18 ms, respectively. These results suggest the proposed KMS is usable and operationally responsive for day-to-day MSME use. Limitations include the single-site scope, small sample, and short evaluation window; future work will extend to multi-site validation, longer observation periods, and additional objective service indicators.

Fraud Prediction Model on Premium Cosmetics Transactions Using Deep Learning: A Long Short-Term Memory (LSTM) Approach

Nandita Sekar Sukma Dewi, Aprilisa Arum Sari — Wed, 28 Jan 2026 13:52:40 +0800

The rapid growth of the premium cosmetics industry has significantly increased online and offline transactions, but also heightened the risk of fraud. Traditional detection approaches often fail to capture dynamic patterns. This study proposes a fraud prediction model using Long Short-Term Memory (LSTM), a deep learning architecture suitable for sequential transaction data. Unlike previous studies that mainly focus on banking and general e-commerce fraud, this research specifically addresses premium cosmetics transactions, a domain with limited exploration. The dataset consists of 2,133 transactions with 16 features covering demographics, transaction details, and technical attributes. After preprocessing (cleaning, normalization, categorical encoding, and sequential arrangement), the LSTM model was trained and validated (70-15-15 split), achieving 94.2% accuracy, 91.5% precision, 89.7% recall, 90.6% F1-score, and 0.95 AUC. These results highlight the novelty and effectiveness of LSTM in detecting fraudulent patterns in the premium cosmetics sector, offering practical implications for enhancing security and trust in high-value transactions.

Automated Hyperparameter Optimization Using Optuna for EfficientNet-Based Medical Image Classification

Windra Swastika, David Yusaku Setiyono, Bita Parga Zen — Wed, 28 Jan 2026 15:12:32 +0800

Manual hyperparameter tuning remains a significant bottleneck in developing robust deep learning models for medical applications. This study presents a comprehensive analysis of Optuna's Tree-structured Parzen Estimator (TPE) for automated hyperparameter optimization of EfficientNet-B2 architecture in Acute Lymphoblastic Leukemia (ALL) cell classification. Using the C-NMC dataset comprising 10,661 training and 1,867 test images, we conducted 20 optimization trials with architecture-specific search spaces targeting learning rate (1×10⁻⁵ to 1×10⁻²), dropout rates (0.1-0.5), weight decay (1×10⁻⁶ to 1×10⁻²), and hidden layer sizes (256-1024 neurons). Results demonstrate that learning rate dominates optimization importance (55%) followed by dropout regularization (34%). The framework achieved optimal configuration with 96.86% validation accuracy, reducing manual tuning time by approximately 90% while maintaining its performance (86.72% test accuracy, 0.92 AUC-ROC). Statistical analysis across multiple runs shows consistent performance with coefficient of variation of 1.96%, validating the reliability of TPE-based optimization for medical imaging applications.

The Role of Hashtags in Driving Instagram Engagement: An Analysis of Indonesian Independence Day Content

Wed, 28 Jan 2026 15:20:47 +0800

This study investigates the intricate relationship between Instagram hashtag patterns and user engagement trends. Utilizing a quantitative research design with a descriptive-correlational approach, the research systematically analyzed publicly available Instagram post data collected within a specific timeframe, focusing on content related to the Indonesian Independence Day celebration. The methodology involved rigorous URL cleaning, web scraping for hashtags and likes data, comprehensive data cleaning, and hashtag normalization. Subsequently, descriptive statistics were used to characterize hashtag usage, followed by correlation analysis and multiple linear regression to determine the statistical relationship and predictive power of hashtag patterns on engagement metrics. Initial findings indicate a weak or non-existent linear correlation between the sheer number of hashtags and likes (Pearson: 0.0345), suggesting that more hashtags do not automatically guarantee increased engagement. However, specific hashtags like #hutri80 and #17agustus demonstrate higher average likes, highlighting the importance of relevance and specificity over quantity. The analysis also revealed that a few viral posts significantly skew average engagement for certain popular hashtags, indicating the strong influence of outliers. Overall, the study concludes that while hashtags are crucial for discoverability, content quality, relevance, and other contextual factors likely play a more significant role in driving Instagram user engagement.

Design and Implementation of a Game-Based Learning System for Slow Learner Students in Visual Communication Design Department

Evelyne Henny Lukitasari, Farid Fitriadi, Yunita Primasanti — Wed, 28 Jan 2026 15:45:37 +0800

An inclusive learning approach is essential in vocational education, especially for slow learners who need more focused, gradual, and visual experience-based learning strategies. The complexity of the material in Visual Communication Design (DKV) often makes it hard for them to grasp design concepts, visual principles, and creativity tasks that require significant cognitive effort. These challenges led to the development of game-based learning (GBL) as an alternative method aimed at increasing engagement, retention, and learning effectiveness. This study designs and implements a GBL system for slow learner DKV students at SMK N Surakarta using a design and development research approach. This includes needs analysis, design, development, and implementation testing. The system incorporates the principles of visual scaffolding, simple narratives, task-based interactions, and gamification through a point system, gradual levels, and instant feedback. Testing shows that GBL helps improve learning focus, strengthens understanding of basic design concepts, and motivates students to complete visual tasks. Teachers observed an increase in independence and self-confidence in students after they interacted with the game. These findings confirm that GBL is an effective way to create a more inclusive learning environment in vocational design education. They also open up opportunities for the development of more adaptable and personalized digital learning systems in future research.

Determinant Factors of SeaBank Application Success for Digital Payments Using Extended Technology Acceptance Model

Zahra Nur Anisya, Adnan Purwanto, Singgih Briandoko, Wika Purbasari — Wed, 28 Jan 2026 16:17:47 +0800

This study explores the crucial determinants affecting the successful adoption of SeaBank's digital payment application through Extended Technology Acceptance Model (ETAM), incorporating trust as an additional external variable. As one of Indonesia's fastest-growing digital banks, SeaBank provides distinctive features but improve user convenience and support the transition to cashless transaction. Using a quantitative methodology, were gathered from 100 active SeaBank users through an online survey. The relationship between trust, perceived ease of use, perceived usefulness, attitude, and intention to accept e-payment were analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM) in SmartPLS 4. The analysis substantiates that trust considerably mitigates user concerns regarding data privacy and security. Additionally, both perceived ease of use and perceived usefulness positively shape user attitudes, which subsequently drive the intention to accept e-payment services. This study validates the relevance and suitability of the ETAM framework in the context of digital banking and underscores the necessity of building user trust, ensuring ease of use, and demonstrating clear benefits in platforms such as SeaBank.

Enhancing Decision-Making in Local Government through K-Means Clustering of Structural Official’s Performance

Wed, 28 Jan 2026 00:00:00 +0800

Employee performance evaluation is a critical process in public sector management. However, in Karanganyar Regency, this process has been traditionally conducted on an individual basis, leading to inefficiencies and a lack of actionable insights. This study addresses the gap by applying the K-Means clustering algorithm to categorize the performance of structural officials based on 2021 Employee Performance Target (SKP) data. Key performance indicators include SKP Value, Service Orientation, Commitment, Cooperation, Leadership, and Work Initiative. Using RapidMiner, the data was clustered into three categories: “very good,” “good,” and “satisfactory.” The clustering quality was validated using the Davies-Bouldin Index (DBI), achieving an optimal value of 0.113, which indicates high intra-cluster similarity. The results provide a data- driven foundation for more efficient performance assessments, aiding decision-making in promotions and personnel management. This study demonstrates the potential of machine learning, specifically K-Means clustering, in improving administrative processes and strategic decision-making within local government.

Automated Detection of Foot Tumor: A Machine Learning Approach Leveraging GLCM Texture Analysis

Asyafa Ditra Al Hauna, Raphon Galuh Candraningtyas, Yit Hong Choo — Thu, 29 Jan 2026 12:14:08 +0800

Foot tumors are rare but diagnostically challenging due to overlapping symptoms with benign conditions. Automated image-based detection can aid early identification and reduce misdiagnosis. This study explores the use of GLCM-based feature extraction to classify foot magnetic resonance imaging, focusing on the presence or absence of tumors. The features were classified using logistic regression, decision tree, and random forest. Model performance was evaluated under a five-fold cross-validation framework with scaled features. Experimental results demonstrated strong classification performance, with all models achieving scores between 0.97 and 1.0 across defined metrics. Correlation analysis further revealed that homogeneity, energy, and angular second moment (ASM) had negative associations with the target, while other features showed positive correlations. These findings provide evidence that classical machine learning models, supported by feature engineering, are effective for the detection of foot tumors in absence and presence.

Analysis of the User Experience of Auto-Battler Magic Chess: Go Go Game using Game Experience Questionnaire (GEQ)

Wed, 28 Jan 2026 16:32:54 +0800

This study aims to analyze the user experience (User Experience/UX) in the mobile auto-battler game Magic Chess: Go Go using the Game Experience Questionnaire (GEQ) instrument of the Post-Game module. The study used a descriptive quantitative approach with 100 respondents obtained through the distribution of an online questionnaire. Data were analyzed using descriptive statistics with average calculations, frequency distribution, and Respondent Achievement Rate (TCR). The results showed that the Positive Experience dimension obtained the highest average score (4.14; TCR of 83%) in the Very Good category, indicating that the majority of players feel satisfied, happy, and entertained after playing. Meanwhile, the Negative Experience dimension (2.45; TCR 49%), Tiredness (2.63; TCR 53%), and Returning to Reality (2.72; TCR of 54%) is in the Sufficient category, suggesting that players experienced mild frustration, cognitive fatigue, reduced focus, and minor difficulty transitioning back to daily activities after gameplay. These findings imply that although the game provides strong positive emotional engagement, developers should consider improving balance mechanisms, reducing repetitive gameplay load, and managing in-game intensity to minimize negative psychological effects and maintain long- term player retention.

Design and Implementation of a Network-Based Real Time Monitoring System for Smart Incubators Using IoT

Diyah Ruswanti, Muhammad Ikhsan Hidayat, Jallu Satrio Murdowo, Mukhlis Al Hakim — Thu, 29 Jan 2026 11:46:51 +0800

Temperature setting is very important during the process of hatching chicken eggs, and it's something that poultry farmers worry about a lot. When the temperature isn't steady, it can lead to fewer eggs hatching and longer time needed for the eggs to develop. Regular incubators can keep the temperature somewhat controlled, but they don't always respond quickly to sudden changes like when there's a power cut or a technical problem. This study introduces a system that uses IoT technology to monitor incubation conditions in real time. The system uses a DHT22 sensor to measure temperature and humidity, a relay module to control the heating automatically, and an ESP32 microcontroller that connects to the internet. Result of the data is sent to server using the MQTT protocol, which makes it easy and fast to share the information. The system also includes an API so farmers can view the data on a web page or through apps Telegram. If the temperature goes out of the safe range, the system sends an alert so farmers can fix the problem right away. The experiments show that this system makes monitoring easier, increases the number of eggs that successfully hatch, and gives farmers a reliable, quick, and easy way to manage their incubation process from a distance.

Sentiment Classification of FatSecret Application Reviews with Machine Learning Models

Mayang Gumelar, Farid Fitriyadi — Wed, 28 Jan 2026 16:27:58 +0800

In the current digital era, mobile applications have become an indispensable part of daily life, leading to a surge in user reviews as invaluable repositories of opinions. Health and fitness applications, such as FatSecret, generate millions of reviews rich with insights. However, specific sentiment analysis on FatSecret reviews using a structured Machine Learning (ML) approach remains limited. This study presents a comprehensive approach for sentiment classification of FatSecret application reviews using ML models. We collected Indonesian-language reviews from the Google Play Store, performed extensive data pre-processing (case folding, tokenization, filtering, normalization), and extracted features using Term Frequency-Inverse Document Frequency (TF-IDF) and Bag of Words (BoW). Subsequently, we trained and evaluated five distinct sentiment classification algorithms: Random Forest, Decision Tree, Logistic Regression, SVM, and XGBoost, utilizing the StratifiedKFold method for automatic splitting in training and validation. Evaluation metrics include accuracy, precision, recall, and F1-score. The results of this research are expected to provide deep insights into user perceptions of FatSecret, identify favored and criticized features, and offer a replicable methodological framework for sentiment analysis of other applications in the future.

A Lexicon-Based VADER Approach for Aspect-Based Sentiment Analysis in the Indonesian Language

Muhammad Arsy Al Banna, Siti Khomsah — Thu, 29 Jan 2026 12:07:56 +0800

Aspect-Based Sentiment Analysis (ABSA) provides detailed insights into customer opinions by identifying specific aspects—such as product, service, and management—in textual reviews and analyzing the sentiment toward each aspect. Unlike general sentiment analysis, ABSA reveals which dimensions of customer experience require improvement. However, applying ABSA in low-resource languages like Indonesian is challenging due to limited annotated dataset, sentiment lexicon, and pre-trained model, which often reduce the accuracy of machine learning or deep learning approaches. This study employs the Valence Aware Dictionary for Sentiment Reasoning (VADER), a lexicon-based algorithm effective in analyzing short, informal, and mixed-language texts, such as online reviews. VADER enables reliable sentiment scoring without large labeled datasets, making it suitable for Indonesian-language analysis. A total of 8,438 Google Maps reviews from 2016 to 2025 were analyzed to observe sentiment trends over time. Keywords were developed for three main aspects: product (1,112 words), service (468 words), and management (666 words). Results show that most reviews express positive sentiment (85.4%), followed by neutral (9.9%) and negative (4.6%). The product aspect was most discussed (7,839 reviews), followed by management (4,608) and service (4,589). In conclusion, VADER-based ABSA can effectively analyze customer sentiment in low-resource languages, providing actionable insights to guide restaurant service improvements. The lack of VADER are an obstacle in handling nuance in the Indonesian language and many keywords cannot be extracted by VADER. Futher, method development is needed for more precise aspect extraction.

Evaluation of Village E-Government in Banyumas Regency Using the UN E-Government Development Index

Syefulloh Syefulloh, Adnan Purwanto, Evi Martiani, Tarwoto Tarwoto — Thu, 29 Jan 2026 11:56:07 +0800

In modern governance practices, digital transformation has become a fundamental foundation for improving the quality of public services, transparency, and participation. This study aims to assess the maturity level of e-government implementation across all villages in Banyumas Regency. Using a quantitative approach and census method, this research employs an evaluation model consisting of eleven parameters. The first stage involved determining the total number of villages in Banyumas Regency (297 villages), followed by direct observation to assess the availability of official village websites. The findings show a very high adoption rate, with 285 villages (96.0%) maintaining an active online presence. These 285 websites were then classified into two categories—278 Independent and 7 Integrated—and further evaluated using ten additional parameters adapted from the four-stage UN E-Government Development Index (UN-EGDI), covering Emerging, Enhanced, Transactional, and Connected Presence. The results provide a comprehensive mapping of the two models of village e-government implementation and highlight significant disparities between website availability and functional service delivery. Although adoption is nearly universal, the functional maturity of village websites remains relatively low, indicating a clear gap in digital service readiness. The insights from this study are expected to support strategic policy formulation aimed at promoting more equitable and sustainable digital transformation at the village level. As future work, this research will be extended to the provincial scope by evaluating e-government maturity across districts and cities in Central Java Province.

Deep Learning-Based Herbal Plant Classification Using Leaf Shape and Pattern: The UII Botanical Leaf Dataset

Thu, 29 Jan 2026 12:22:19 +0800

Herbal plants play a crucial role in healthcare and are widely used as traditional medicines. However, identifying herbal species remains a major challenge due to morphological similarities, particularly in leaf shape and texture. This study aims to develop an intelligent classification system for Indonesian herbal plants based on leaf image analysis using artificial intelligence (AI) and digital image processing techniques. A localized dataset of 47 herbal species collected from Botanical SmartPark SMA UII was used to train a deep learning model employing the MobileNetV2 architecture through transfer learning. The proposed model achieved an average accuracy of 96.6% on the testing dataset, demonstrating high reliability in recognizing species with complex visual variations. The trained model was then implemented into an Android-based application called HERBfull Botanical SmartPark, enabling real-time plant identification and interactive access to botanical information. The system successfully enhances efficiency, accessibility, and educational value in the identification of local herbal species. This research contributes to the advancement of AI applications in botanical education, promoting digital literacy, biodiversity conservation, and the integration of smart technology into sustainable environmental learning platforms.

Performance Comparison of Breast Cancer Classification Methods: Naive Bayes vs. Support Vector Machine

Tutus Pandam Pradipta, Sri Huning Anwariningsih — Thu, 29 Jan 2026 00:00:00 +0800

Breast cancer is a global health issue where early detection and accurate diagnosis play a key role in improving patients' chances of successful recovery. Despite their widespread use and proven effectiveness, traditional diagnostic methods have limitations that have prompted the development of computational approaches. Machine learning is one such approach. Numerous prior studies have investigated various algorithms, including Naive Bayes and Support Vector Machine (SVM), for breast cancer classification; however, research directly comparing their performance on the same dataset is still limited. This study evaluates the efficacy of Naive Bayes and SVM methods for classifying breast cancer diagnoses as benign or malignant using the publicly available Wisconsin Diagnostic Breast Cancer (WDBC) dataset. The research stages include data collection, preprocessing, splitting the dataset into training and test sets at 70% to 30%, standardizing features for the SVM model, applying both algorithms, and evaluating performance using metrics such as accuracy, precision, recall, and F1-score. The test results indicate that the SVM algorithm achieved an accuracy of 98.25%, precision of 100%, recall of 95%, F1-score of 98%, and MCC of 0.96. Conversely, the Naive Bayes algorithm achieved 94.15% accuracy, 94% precision, 91% recall, a 93% F1-score, and 0.88 MCC. The comparison results indicate that SVM outperforms Naive Bayes on this dataset, especially in reducing false- positive and false-negative rates. This research is expected to serve as a valuable resource for medical professionals and researchers seeking to select the appropriate machine learning algorithm for early breast cancer detection.

Forest Fire Detection Leveraging Hybrid Convolutional-Recurrent Models

Raphon Galuh Candraningtyas, Asyafa Ditra Al Hauna, Mohamad Nassar — Thu, 29 Jan 2026 00:00:00 +0800

Forest fires pose serious environmental and economic risks across tropical, temperate, and boreal regions. Traditional detection methods are often limited in accuracy and adaptability, motivating the use of deep learning for automated solutions. While Convolutional Neural Networks (CNNs) have shown promise, fewer studies have systematically examined hybrid models combining CNN feature extraction with recurrent layers such as Recurrent Neural Networks (RNN), Long Short-Term Memory (LSTM), and Gated Recurrent Units (GRU). This study compares CNN-MLP, CNN-RNN, CNN-LSTM, and CNN-GRU architectures on a public forest fire dataset, evaluating classification performance and computational efficiency. Results show that CNN-GRU offers the best trade-off, closely matching CNN-MLP in accuracy while requiring fewer resources. CNN-LSTM provides stable performance, whereas CNN-RNN underperforms and needs refinement. Computational analysis further indicates that CNN-MLP is the the most resource intensive models with over 1 millions parameter. These findings highlight CNN-GRU as a strong candidate for real-time forest fire detection, balancing accuracy and efficiency, and suggest future exploration of adaptive thresholds and transformer-based approaches.

Palm Fruit Ripeness and Quality Detection System Using YOLOv11

Queenta Paradissa Ramadhani, Favian Dewanta — Thu, 29 Jan 2026 12:40:59 +0800

The ripeness level of palm fruit is a crucial factor that determines the quality and efficiency of palm oil production. Manual ripeness assessment is often subjective, inconsistent, and time-consuming, creating the need for an automated solution. Therefore, an automated approach using computer vision is needed to ensure efficiency and consistency. To address this need, this study implements the YOLOv11 deep learning model to classify palm fruit into four categories (unripe, underripe, ripe, and overripe). The dataset, obtained from Roboflow, consists of 800 annotated images evenly distributed across the four classes. Data preparation included resizing images to 640×640 pixels and applying augmentation techniques to improve model generalization. The model was trained for 100 epochs on google colab with GPU L4 acceleration. Evaluation results demonstrate high performance with [email protected] of 97.4% and [email protected]:0.95 of 94.1%, alongside precision of 94.7% and recall of 90.6%. The best performance was achieved on the unripe and underripe classes, while the ripe category showed relatively lower accuracy due to visual similarities with adjacent classes. These findings confirm that YOLOv11 is an effective and efficient approach for automatic palm fruit ripeness detection, offering potential benefits for harvesting optimization and supporting smart farming practices.

Comparative Study of CNN, Vision Transformer, and Hybrid CNN–ViT Models for Indonesian Batik Pattern Classification

Naufal El Kamil Aditya Pratama Rahman, Akmelia Zahara, Bintang Yudhistira — Thu, 29 Jan 2026 12:49:58 +0800

Batik is an Indonesian cultural heritage with unique visual characteristics and deep philosophical value. The complexity of motifs, color variations, and geometric details make batik classification an interesting challenge in the field of computer vision. This study conducted a comparative study between three deep learning approaches for classifying Indonesian batik motifs using Convolutional Neural Network (CNN), Vision Transformer (ViT), and a hybrid CNN–ViT model. The dataset used includes more than 3,000 batik images from various regions in Indonesia, with a variety of motifs such as Yogyakarta Kawung, Aceh, Ceplok, and Megamendung.Each model was trained with uniform parameters and augmentations to ensure fair evaluation, resulting in CNN accuracy of 94.43% F1-macro 93.45%, ViT accuracy of 91.55% F1-macro 89.78%, and Hybrid CNN-ViT accuracy of 94.04% F1-macro 92.91%. This is reinforced by the combination of modules (EfficientNet-B2 + CBAM + ArcFace) that can improve model performance furthermore, This study contributes to the development of an automated batik classification system and supports cultural preservation through artificial intelligence- based digitization.

Automated Detection of MRONJ Lesions in Panoramic Dental X-rays Using Candidate Region Identification and Semantic Segmentation

Thu, 29 Jan 2026 13:24:56 +0800

Medication-related osteonecrosis of the jaw (MRONJ) is a severe adverse effect associated with the administration of bone-modifying agents, such as bisphosphonates (BP) and denosumab (Dmab), and angiogenesis inhibitors. Despite the advancements in therapeutic agents, the incidence of MRONJ has increased, as medication remains a primary risk factor. In most cases, MRONJ is diagnosed at an advanced stage, where portions of the jawbone become exposed in the oral cavity, interfering with both primary disease management and MRONJ treatment. Therefore, early detection and treatment prior to progression are critical for improving patient outcomes and reducing treatment complexity. In Japan, the low penetration of dental CT limits the feasibility of 3D diagnostic imaging in routine practice in dental clinics. Therefore, this study proposes a diagnostic method that relies solely on panoramic X-ray images to automatically predict MRONJ lesions. The proposed method first performs pre-processing to extract the mouth region, and then compares two approaches for MRONJ lesion segmentation. The first approach subdivides the mouth region into patches and utilizes patch-based classification to identify candidate regions before MRNOJ lesion segmentation. The second approach employs the masked vision transformer (Masked-ViT) to estimate the probability of MRONJ lesion presence across the image, and then segmentation is applied to high probability areas. On our panoramic X-ray image dataset consisting of 118 MRONJ patients, the patch-based method achieved a maximum Dice Similarity Coefficient (DSC) of 0.70, outperforming the method using Masked-ViT. Although promising, further enhancements are necessary to meet the requirements for clinical use.

Integrating Supervisor Access Using Hybrid RBAC-ABAC In A Web-Based Research Permit Information System

Santyana Rahmawati, Hanifah Permatasari, Intan Okaviani — Thu, 29 Jan 2026 00:00:00 +0800

The increasing demand for research activities in hospitas requires a secure, reliable and efficient information system to manage research permit applications. In many healthcare institutions, supervisory teams, play a crucial role in monitoring research activities to ensure compliance with institusional policies and ethical standards. This study presents the integration of supervisory team access into the existing web-based Research Permit Information System at Dr. Moewardi General Hospital. The integration is designed to enable the supervisory team to directly access and review research data through the system with access control aligned by hospital’s organizational hierarchy and regulations. To enhancing security and handling some access scenarios, a hybrid access control model combining Role-Based Access Control (RBAC) and Attribute-Based Access Control (ABAC) is implemented. RBAC is used to define role-specific permissions for different supervisory levels, ensuring consistent enforcement of access boundaries. ABAC complements this by allowing more granular, attribute-driven policies that improve adaptibility to dynamic and context-specific access requirements. The proposed hybrid model strengthens system security and increase flexibility in access management for various supervisory role. This approach demonstrates a practical and scalable solution for integrating multiple access control mechanism in a healthcare research context.

Performance Portability, Reliability, Usability, and Maintainability of PT. ASDP Indonesia Ferry’s E- Procurement Website Based on ISO/IEC 25010 Standards)

Dwi Retnoningsih, Astri Charolina, Afif Faris Hudaifi — Thu, 29 Jan 2026 14:27:08 +0800

Website quality influences customer satisfaction, building a positive reputation, time, cost, and process efficiency. The importance of a website for a company’s professionalism means that website quality assurance is crucial. A comprehensive software quality assurance standard is ISO/IEC 25010. This international standard serves as the basis for measuring the quality of web-based software. The software product quality model includes eight characteristics: functional suitability, performance efficiency, compatibility, security, usability, reliability, portability, and maintainability. The E-Procurement website is the operational medium for PT. ASDP Indonesia Ferry’s digital business. The company’s efforts to ensure website quality have been measured using the ISO/IEC 25010 standard, which is limited to four characteristics: functional suitability, performance efficiency, compatibility, and security. This study proposes measuring website performance by examining four other characteristics: usability, reliability, portability, and maintainability. Performance measurement was carried out using the Mean Opinion Score (MOS) method and the WAPT test tool. The results showed that the usability characteristic reached 83%, meaning the system is simple enough to be used and understood by users. The website’s reliability proved stable and capable of handling access loads without failure, according to WAPT testing. In terms of maintainability, the use of the component-based Angular framework simplifies system development and maintenance. Portability testing showed the website experienced no significant issues across browsers and operating systems. These four characteristics demonstrate that the E-Procurement website is easy to use, highly reliable, easy to maintain, and compatible with various platforms.

Applying Consistent Fuzzy Preference Relation in Weighting Software Effort Estimation Criteria

Ika Indah Lestari, Adnan Purwanto, Sulistiyasni Sulistiyasni, Singgih Briandoko — Thu, 29 Jan 2026 15:20:48 +0800

Software effort estimation (SEE) is a critical process in project planning, as it determines budget allocation, resource management, and timeline accuracy. The weighting of estimation criteria significantly influences the reliability of the estimation model. This study aims to determine the weights of SEE criteria using a fuzzy logic approach, specifically the Consistent Fuzzy Preference Relation (CFPR) method. As a Multi-Criteria Decision Making (MCDM) technique, CFPR offers an efficient mechanism for extracting consistent expert preferences by requiring only n-1 pairwise comparisons from n criteria, making it suitable for rapid weighting calculations. The study evaluates four main attributes: Product, Computer, Personnel, and Project. Expert assessments were conducted using crisp numbers on a 1-9 scale. The results show the following attribute weights: Product (0.372), Computer (0.275), Personnel (0.231), and Project (0.122). Furthermore, the top three ranked cost drivers are Required Reliability (0.1674), Product Complexity (0.1384), and Execution Time Constraint (0.1050). Conversely, the lowest weights were assigned to Programming Language Experience (0.0270), Virtual Machine Experience (0.0296), and Required Development Schedule (0.0303). The integration of CFPR into SEE models produces a stable and interpretable weight distribution, thereby enhancing the accuracy of effort estimation.

Optimization of Random Forest Model via GridSearchCV for Hoax News Detection

Lutvi Riyandari, Singgih Setia Andiko, Siti Delimasari, Singgih Briandoko — Thu, 29 Jan 2026 00:00:00 +0800

In this time of fast digital information growth, information sources can be helpful or harmful. The internet makes it easier for people to find information, but it also makes it easier for fake news and hoaxes to spread quickly and widely. This work seeks to combat the dissemination of false news in the digital age by employing text categorization through the Random Forest algorithm, coupled with hyperparameter optimization via Grid SearchCV.The dataset comprises both hoax and authentic news from Indonesia, subjected to various steps including text processing (case folding, tokenization, and stopword elimination) and feature weighting via TF-IDF.The study’s results reveal that the Random Forest model does an impressive job of telling the difference between fake and real news when tested using a confusion matrix. The confusion matrix shows that the model works better after hyperparameter tweaking with GridSearchCV. This is shown by the fact that the number of accurate predictions (TN and TP) goes up and the number of wrong predictions (FP and FN) goes down. The evaluation measures (accuracy, recall, precision, and F1-Score) also demonstrate significant improvements, increasing from 96% to 97%.

Comparison of Ensemble Learning Methods on the IoT-23 Dataset

Syakira Az Zahra, Kurnia Anggriani, Agus Susanto — Thu, 29 Jan 2026 16:04:14 +0800

The Internet of Things (IoT) has provided numerous benefits across various sectors, but it also poses significant challenges in cybersecurity, particularly malware threats. Malware on IoT devices has the potential to damage systems, steal data, and disrupt network performance. Previous research has shown that the Na ̈ıve Bayes algorithm produces a low accuracy of 0.24, increasing slightly to 0.35 when combined with AdaBoost, and reaching 0.99 when combined with XGBoost using the soft voting method. However, there is still room to explore other ensemble learning methods to obtain more stable results. This research focuses on the application of an alternative ensemble learning method, namely stacking, using the IoT-23 dataset with reference to the CRISP-DM framework. The results show that the stacking method can significantly improve malware detection accuracy from 0.35 to 0.72, thus proving superior to soft voting and can be an effective approach in improving malware detection performance in IoT networks.

The Structural Interaction between Teachers’ Collaboration and Inquiry-Based Learning Networks: Effects on the Implementation of Inquiry-Based Learning

Masanori Fukui, Hari Widi Utomo, Andi Prademon Yunus, Asyafa Ditra Al Hauna — Thu, 29 Jan 2026 00:00:00 +0800

The purpose of this study is to clarify the structural relationship between teacher collaboration and Inquiry-Based Learning (IBL) networks in the diffusion process of IBL among Japanese teachers (N = 650). Teacher collaboration was measured using three subscales: teacher collaboration among colleagues (CF1), leadership by administrators (CF2), and partnerships with communities and parents (CF3). The analysis revealed high correlations among CF1–CF3 (r = .62-.69) and moderate correlations with the IBL Network (r = .31-.41), indicating the interconnection between intra-school and extra-school collaborative cultures and IBL networks. Furthermore, a Seemingly Unrelated Regression (SUR) model demonstrated a statistically significant reciprocal facilitation relationship between collaboration and the IBL network. These results suggest that the diffusion of IBL is mutually promoted by strong internal interactions (collaboration within schools), external network linkages (diffusion across contexts), and the formation of inquiry-related teacher networks.

Integration of YOLOv11 and Convolutional Neural Network in a Deep Learning Approach for Coffee Bean Defect Detection and Classification

Fri, 30 Jan 2026 14:23:30 +0800

The coffee industry is a strategic commodity that significantly contributes to the global and national economy. Coffee bean quality strongly influences flavor and market value, while defective beans—such as broken, moldy, or quaker beans—can reduce overall quality. Manual sorting methods, still widely used by farmers and small-scale producers, are time-consuming, inefficient, and prone to human error. This study proposes an automated deep learning–based system for detecting and classifying defective coffee beans by integrating YOLO for object detection and EfficientNetV2 as the classifier. A dataset of 5,636 coffee bean images from multiple sources was used. The system was evaluated through black box testing to ensure the functionality of the web interface and performance testing using a confusion matrix. Results show that YOLOv11 achieved an [email protected] of 98.83%, while EfficientNetV2 obtained a test accuracy of 93.81%. The proposed system demonstrates strong potential to improve coffee sorting by providing a faster, more accurate, and efficient alternative to manual methods.

Prioritization Analysis of Instagram Digital Marketing Strategy HMBD Telkom Purwokerto with SWOT and AHP Methods

Thu, 29 Jan 2026 16:28:12 +0800

The digital era has changed the paradigm of student organization communication, which needs to optimize digital marketing strategies to increase visibility and engagement. Instagram, with 90.18 million users in Indonesia, is a potential strategic platform but requires a systematic approach in determining strategic priorities. This study aims to determine the optimal digital marketing strategy priorities for the Instagram account @hmbd.telkompurwokerto through an integrated approach combining SWOT Analysis and the Analytical Hierarchy Process (AHP). The research method employs a mixed-method approach with a single-case study design, involving in-depth observation, structured interviews, and focus group discussions for SWOT analysis, as well as pairwise comparisons with 4–5 hmbd committee members for AHP implementation. The research results are expected to identify the main strengths, namely institutional credibility and the quality of educational content; the main weaknesses, namely budget constraints and inconsistent posting schedules; the greatest opportunities in stakeholder collaboration and new Instagram features; and the main threats, namely dynamic algorithms and content saturation. The criteria “Content Quality & Relevance” and “Audience Engagement Strategy” are predicted to have the highest weight of importance in AHP. The research contributes theoretically through a SWOT-AHP integration model for student organization digital marketing and practically through an implementable strategic roadmap that can be adapted by other student organizations.

Selection Strategy for Handling Deadstock Products using the Analytical Hierarchy Process (AHP) and Expected Monetary Value (EMV) Method

Thu, 29 Jan 2026 00:00:00 +0800

The accumulation of deadstock on Network Terminal Equipment (NTE) products in the telecommunications company's warehouse has resulted in increased storage costs and financial losses for the Company. This study aims to develop an optimal deadstock-handling selection strategy by integrating the Analytical Hierarchy Process (AHP) and the Expected Monetary Value (EMV). The AHP method is used to determine the weight of the decision criteria (cost, implementation time, and ease of implementation). At the same time, EMV is calculated using a Monte Carlo simulation to evaluate the monetary value of each of eight handling alternatives (re-layout, FIFO strategy, purchase forecast, stock opname, product discount, resale to supplier, sale to Marketplace, and sale with bundling strategy). Data were collected through expert interviews and historical deadstock data from the telecommunications company's warehouse. The analysis shows that the AHP-EMV combination can recommend the best strategy by considering both financial and non-financial factors. The EMV simulation revealed that sales with a bundling strategy provide the highest monetary value of IDR. 288,572,250 compared to other alternatives. This research offers practical contributions in the form of a data-based decision-making framework for deadstock management and policy recommendations for internet service providers.

Application of Distribution Requirement Planning in Optimizing Packaged Drinking Water Distribution to Mitigate Lost Sales

Thu, 29 Jan 2026 17:01:35 +0800

The bottled water industry in Indonesia has experienced a positive growth trend due to increasing public awareness of the importance of hydration and a healthy lifestyle. However, DC Formula Paluta in Padang Lawas Utara, North Sumatra, faced a lost-sales challenge of 7% in 2023, exceeding the typical tolerance level in the FMCG industry, which ranges from 3–5%. The purpose of this study is to identify the causes of lost sales and propose solutions through demand forecasting, determining the optimal order quantity, and planning a more efficient distribution using the DRP Worksheet and fishbone diagram. Historical data analysis from January to December 2023 using Pom QM software shows that the Regression/Trend Analysis and Multiplicative Decomposition methods have the lowest Mean Absolute Deviation (MAD) values. Therefore, these two methods were selected for forecasting the demand for gallon, bottle, and cup products in the coming year. The implementation of the Economic Order Quantity (EOQ) method successfully reduced the ordering frequency from 190 times to 124 times per year, saving costs from IDR 147,440,000 to IDR 96,224,000, or approximately 35%. The application of Distribution Requirement Planning (DRP) improved the efficiency of distribution planning and inventory management, ensuring product availability according to market demand. The combination of forecasting methods, EOQ, DRP, and fishbone diagram analysis is expected to reduce mismatches between inventory and demand, improve operational and distribution management efficiency, and directly lower the level of lost sales occurring at DC Formula Paluta.

A Systematic VDI 2221 Methodology for Piezoelectric Energy Harvesting in Ergonomic Lumbar-Support Wearables Product

Thu, 29 Jan 2026 16:54:54 +0800

The need for sustainable, portable renewable energy sources is increasingly crucial, especially for human activities in remote areas with minimal access to electricity. The research aims to develop Piezo-Powered Ergo-Lumbar Support Device prototype, an ergonomic lumbar support cushion-backpack integrated with a portable renewable energy source through a piezoelectric energy harvesting system. A combination of VDI 2221 and Human-Centered Design (HCD) methods was applied to simultaneously optimize the technical and ergonomic needs of users. A technical-economic evaluation was used to determine the best design solution concept. Vibration simulation test in Solidwork software confirmed that the product can accommodate vibrations up to an average of 1218 Hz, which is required to activate 32 piezoelectric elements arranged in parallel to optimally convert kinetic energy into electrical energy. This research results in a final prototype with dimensions of 38.5 x 26.5 x 10 cm and a weight of 475 grams with a power storage capacity of 1200 mAh. Evaluation conducted on 43 respondents proved that the product increases comfort and improves body posture (88.4% of respondents agreed), and the feature of generating renewable energy independently is considered innovative and useful. This product offers a promising integrated ergonomic-energy solution for sustainable energy innovation.

Green Innovation Product for Sustainable Waste Management

Centaury Harjani, Servatia Mayang Setyowati, Restituta Wening G. Gyarwahyu — Thu, 29 Jan 2026 17:17:36 +0800

The management of glass bottle waste remains a critical environmental challenge in urban centers like Yogyakarta, Indonesia. This study investigates the upcycling of this waste stream into high-value, functional products, such as platters, through a combination of cold working (cutting) and hot working (flameworking), within the Green Innovation Product (GIP) paradigm. An experimental methodology was employed, comparing the efficiency of a small, updraft kiln (2023 experiment) with a larger, downdraft kiln (2025 experiment). Results demonstrated that while utilizing the entire glass bottle (neck, body, and base) enhanced material efficiency, the larger kiln’s stacked firing arrangement was ineffective, as only the top layers melted completely even with extended firing durations up to 120 minutes. The study concludes that scaling production efficiency is contingent not on kiln size alone, but on achieving uniform heat distribution, for which a single-level, flatbed kiln design is proposed. This research confirms the technical feasibility of transforming glass waste into commercially viable products and underscores the necessity of appropriate technology integration to realize the principles of GIP, offering a scalable model for sustainable waste management and local entrepreneurship.

The Implementation of Green Economy in Increasing the Achievement of Sustainable Development Goals (SDGs) in the Tempe Industry in Sanan Malang

Purnomo Purnomo, Novenda Kartika Putrianto, Nanta Sigit, Samgar Samgar — Thu, 29 Jan 2026 00:00:00 +0800

The application of green economy principles is one of the important strategies in encouraging sustainable development, especially in the small and medium industrial sector. This study aims to analyze the level of application of green economy principles in the tempeh industry and its contribution to the achievement of the Sustainable Development Goals (SDGs). The research method used is a descriptive quantitative approach by collecting data through questionnaires to 30 tempeh industry players. The analysis was carried out using descriptive statistics and Pearson correlation tests. The results of the study showed that the level of green economy implementation was in the “adequate” category with an average score of 3.04 on the Likert scale of 1–5. The highest indicator is in the aspect of economic impact on the community, while the lowest aspect is the use of environmentally friendly technology and waste management. Correlation analysis showed a significant positive relationship between the length of business and the level of green economy implementation (r = 0.45; p= ¡ 0.05), which indicates that business experience also influences sustainability awareness and practices. Although the tempeh industry has a positive contribution to the achievement of SDG 8 (decent work and economic growth), its contribution to SDG 12 (responsible consumption and production) and SDG 13 (handling climate change) is still limited. Policy interventions in the form of training, incentives, and adoption of clean technologies are needed to increase synergy between green economy practices and SDGs goals.

Mitigation of Inter-Carrier Interference (ICI) in Mobile DVB-T2 Technology Using Zero Forcing Equalization

Jilan Haidar Rahman, Wahyu Pamungkas, Solichah Larasati — Fri, 30 Jan 2026 12:13:29 +0800

Orthogonal Frequency Division Multiple Access (OFDM) is the transmission mechanism employed in DVB-T2 digital broadcasting systems. However, under mobility conditions, Doppler shift becomes a major factor limiting OFDM performance. The Doppler effect induces Carrier Frequency Offset (CFO), which disrupts the orthogonality among subcarriers and generates Inter-Carrier Interference (ICI). This issue causes frequency mismatches between the transmitter and receiver. In this study, a DVB-T2 system configuration is used with 2k mode, guard interval 268, cyclic prefix 1/4, and 64-QAM modulation. To evaluate mobility effects, simulations were performed with maximum Doppler frequency variations ranging from 2.87 Hz to 40.18 Hz and SNR ranges from 0 dB to 40 dB. On the receiver side, equalization was applied using a Zero Forcing Equalizer. Simulation results indicate that without equalization, the BER remains approximately 0.5 across all Doppler variations. Conversely, with the application of Zero Forcing Equalizer, BER decreases significantly at SNR ≥ 20 dB. This method proves effective in mitigating ICI in DVB-T2 systems under mobility conditions, achieving an average performance improvement of 98.42% with BER approaching zero at low to moderate Doppler frequencies.

Analysis of BTS Selection for Implementation Of Multi Operator Core Network (MOCN): Merger Case Study PT Indosat Ooredoo Tbk With PT Hutchison 3 Indonesia

Ida Tutiek Samsiyah, Wahyu Pamungkas, Alfin Hikmaturrokhman — Fri, 30 Jan 2026 12:30:40 +0800

Multi Operator Core Network (MOCN) is aproject that aims to create a network system core for several telecommunication operators in one system, first implemented in Indonesia by Indosat Ooredoo and Hutchison 3. Before the merger Indosat Ooredoo has 29,503 physical BTS and Hutchison 3 Indonesia has 32,489 physical BTS, a total of 61,992 physical BTS that must be maintained and monitored but they are not yet integrated with each other. This makes the focus on customer service larger and Unintegrated. With MOCN, BTS networks will be able to be integrated and focus on customer service can be maximized. Indonesia currently uses a tower collocation scheme, where in 1 tower there can be more than 1 operator. Similarly, at Indosat Ooredoo and H3I, there are 22,389 BTS units physically located on the same tower, commonly referred to as 'pair collo'. A decision is needed to determine whether Node B should be activated for MOCN or deactivated during the network consolidation process.. Technical analysis by considering BTS capacity, coverage, antenna position, and transmission mode to determine which BTS provides greater benefits to IOH after the network merger. With the network merger, it is hoped that there will be efficiency in network maintenance of 27.6%, as well as maintenance cost efficiency which will be reduced. This research aims to provide strategic and operational solutions for managing the infrastructure, especially related to the selection of operational sites and network optimization. Literature study, case study, and comparative analysis is used in this research. Solution applied includes reconciliation and monitoring post-MOCN performance and ensuring quality the network remains maintained, while maximizing tower rental efficiency.

Culinary Content Delivery Strategies by Micro-Influencer @nyamwithinop in Building Consumer Trust on TikTok Micro-Influencer

Cindy Nasywa Aurora, Hanifah Permatasari, Rizky Betha Nur Ramadhany — Fri, 30 Jan 2026 00:00:00 +0800

The rise of social media has fostered the emergence of the content creator profession, which plays a vital role in the digital economy ecosystem, including the culinary sector. This study aims to analyze the communication strategies employed by the Micro-Influencer @nyamwithinop on TikTok in building credibility and audience trust. The research adopts a descriptive qualitative method with a case study approach, in which data were collected through documentary observation of video content and audience interactions during the period 2020–2025, covering aspects such as presentation style, narrative, visuals, and user responses. The findings reveal that @nyamwithinop’s success in fostering trust is supported by four key communication strategies: (1) understanding the audience by tailoring topics, language, and delivery styles according to audience segments; (2) defining distinct content objectives between premium restaurant reviews and simple culinary experiences; (3) constructing messages through concise and straightforward narratives, complemented by engaging and audience relevant visuals; and (4) selecting appropriate methods and media by leveraging TikTok as the primary platform that aligns with the characteristics of younger generations. These strategies are reinforced by two-way interaction, consistent content posting, and active audience engagement, which collectively enhance both credibility and loyalty. The study underscores that Micro-Influencer’s are effective in shaping consumption decisions through well-crafted communication strategies and provides practical implications for culinary UMKM in designing social media based promotional strategies.

Influencer Marketing as a Catalyst for Sustainable MSME Growth in Indonesia

Arya Budi Sutopo, Sri Huning Anwariningsih — Fri, 30 Jan 2026 00:00:00 +0800

Micro, Small, and Medium Enterprises (MSMEs) constitute 99.9% of Indonesian businesses and employ 97% of the national workforce, yet face persistent challenges in market access and brand visibility. This literature review examines how influencer marketing—particularly through micro-influencers (10K–100K followers) and nano-influencers (<10K followers)—can enable sustainable MSME development in Indonesia. Through comprehensive analysis of academic literature and industry reports, we identify that nano-influencers achieve up to seven times higher engagement rates compared to macro-influencers, making them cost-effective marketing partners for resource-constrained MSMEs. The review integrates three theoretical frameworks: Social Capital Theory explains trust transfer mechanisms in collectivist cultures; Technology Acceptance Model addresses adoption barriers including digital literacy gaps (62% of MSMEs affected); and Stakeholder Theory demonstrates how sustainability-integrated marketing creates multi-dimensional value beyond profit maximization. Key findings reveal that influencer marketing enables MSMEs to compete through authenticity and values alignment rather than advertising budgets, with particular effectiveness on platforms like TikTok where 18% of organic MSME posts exceed 100,000 views. However, systemic barriers persist: infrastructure disparities (33% gap between Java and Eastern Indonesia), financial constraints (67% of MSMEs allocate <2% revenue to marketing), and influencer identification challenges (45% struggle to find authentic partners). The study proposes a five-phase implementation roadmap and ecosystem-level policy interventions targeting government, social media platforms, financial institutions, and industry associations. This research contributes to CENTIVE 2025's mission of advancing sustainable solutions through multidisciplinary approaches, demonstrating how digital marketing innovation can drive inclusive economic development while preserving cultural heritage and environmental resources.

Decision Support System Using the Analytical Hierarchy Process (AHP) Method for Evaluating Marketing Strategy Effectiveness (Case Study: Gressoy Indonesia)

Fri, 30 Jan 2026 00:00:00 +0800

Gressoy Indonesia is a micro business in the food and beverage sector that faces challenges in increasing sales of its products. To assist decision-making in choosing the most effective marketing strategy, this research develops a decision support system using the Analytical Hierarchy Process (AHP) method. This method is used to develop a hierarchical structure based on objectives, evaluation criteria, and marketing strategy alternatives, then weighted through pairwise comparisons. Data was obtained through observation, interviews, and questionnaires with Gressoy management. There are four marketing strategies evaluated, namely promotion through social media, cooperation with partnerships, providing seasonal discounts, and participating in an event. The results of data processing show that the strategy of cooperation with partnerships is the top priority to be evaluated with the highest weighted value of 0.436. Other strategies in order are participating in an event (0.270), giving seasonal discounts (0.187), and promotion through social media (0.106). The criteria that most influenced the decision was ease of implementation, followed by effectiveness in increasing sales, market reach, and cost. These results suggest that strategies that are easy to implement and have a direct impact on sales need to be the main concern. These findings are expected to help MSMEs in developing more effective and targeted marketing strategies to increase product sales.

Data Acquisition System to Support Predictive Maintenance on Soft Laminator Machines in an Electronics Manufacturing Company

Deni Rahmat, Firdaus Firdaus, Dwi Ana Ratna Wati — Fri, 30 Jan 2026 00:00:00 +0800

This study presents the design and implementation of an Internet of Things (IoT)-based data acquisition system for a soft laminator (profile wrapping) machine used in electronic audio device manufacturing. The system aims to enable real-time monitoring of critical process parameters, including heater roll temperature, heater dry zone temperature, and roll spacing, which are essential for maintaining product quality and reducing machine downtime. The proposed system employs an ESP32 microcontroller integrated with DS18B20 temperature sensors and VL53L0X distance sensors, supported by an Ethernet W5500 module for reliable data transmission to a MySQL-based server. A web-based dashboard was developed to visualize sensor data, display alerts, and log historical records. Experimental results show that the system achieved high accuracy, with mean absolute errors of 0.38 °C (0.63%) for heater roll temperature, 0.44 °C (0.73%) for heater dry zone temperature, and 0 mm (0%) for all distance sensors, well within the industrial tolerance of <1%. Additionally, the indicator subsystem—consisting of LEDs and buzzers—responded consistently to simulated fault conditions such as sensor failure and network disconnection. Overall, the developed system demonstrates reliable performance for industrial monitoring applications and offers a foundation for implementing predictive maintenance in manufacturing environments.

Low-Cost IoT-Based Landslide Early Warning System

Fri, 30 Jan 2026 13:26:16 +0800

EWS (Early Warning System) or commonly called an early warning system is an electronic circuit designed to detect systematically that will be monitored online. In this early warning system, it designs a landslide or landslide disaster. This tool uses two microcontrollers including Arduino Nano as a support for data results from both sensors and ESP32-DevKit V1 as a data receiver that has been processed on Arduino Nano and sending data to Blynk Console via 4G LTE WiFi Modem Module with a delay of 48 seconds to get a good signal and a delay of 30 seconds to initiate a connection with the ESP32 DevKit V1 microcontroller. The Slide Potentiometer Sensor which is used to determine the indication of a landslide shift, which has 2 parameters Alert (2-3 cm) and Warning (4-6 cm) with a sensor accuracy level of 98.48%, and has a Soil Moisture Sensor which plays an additional role and indication and humidity conditions in the soil with a sensor accuracy of 90.15%. Thus, both sensors are delivered in an integrated manner with the MQTT Blynk Console application which can provide real-time notifications.

Electrical Energy Consumption Monitoring System for Boarding Rooms Using IoT and Progressive Web Application

Fri, 30 Jan 2026 00:00:00 +0800

, , , ,