Kinetik: Game Technology, Information System, Computer Network, Computing, Electronics, and Control

Intelligent Traffic Management System Using Mask Regions-Convolutional Neural Network

Muhammad Kemal Pasha — 2025-06-13

Urban centers worldwide continue to face challenges in traffic management due to outdated traffic signal infrastructure. This study aims to develop an intelligent traffic management system by implementing the Mask Regions-Convolutional Neural Network (MR-CNN) algorithm for real-time vehicle detection and traffic flow optimization. Utilizing the CRISP-DM framework, this research processes CCTV footage from the Pasteur-Pasopati intersection in Bandung to identify and quantify vehicles dynamically. The proposed system leverages an enhanced Mask R-CNN model with a ResNet-50 FPN backbone to improve detection accuracy. Experimental results demonstrate an 80% vehicle detection accuracy, with a macro-average precision of 0.89, recall of 0.83, and an F1-score of 0.82. These findings highlight the system’s capability to replace conventional fixed-time traffic signals with a more adaptive approach, adjusting green light durations based on real-time traffic density. The proposed solution has significant practical implications for reducing congestion and improving traffic flow efficiency in urban environments.

Integrating ISSM and SCT into the TAM Framework: A Conceptual Model and Empirical Study on E-Government Services

Beny Prasetyo — 2025-06-13

Proposing and developing the right model is necessary to increase the effectiveness and success of e-government service implementation. Combining models highlighting technological aspects and psychological issues can generate satisfaction and improve service quality. This research develops and tests a combination of the Information System Success Model (ISSM), Technology Acceptance Model (TAM), and Social Cognitive Theory (SCT). This research is expected to determine the results of the fit model test of the proposed model developed and empirically test the factors that significantly affect the success of e-government through satisfaction. To validate the conceptual model using PLS-SEM. The type of research conducted is Quantitative research. The sample used to test the model was SiKeren service users in the Jember Regency Government, totaling 260 samples determined using Hair's theory and probability sampling techniques, especially simple random sampling. The results of this study indicate that the proposed model is suitable. The Standardized Root Mean Square Residual (SRMR) value of 0.070 or < 0.08 indicates that the model is considered to be supported by the measured data. The Goodness of Fit (GoF) value is 0.686, indicating a high match between the observed data and the developed model. The model captures the R-Square value of Perceived Ease of Use, Perceived Usefulness, and Satisfaction well by the model, having medium criteria with values of 0.595, 0.724, and 0.606. Of the 16 hypotheses proposed, 12 were accepted and 4 were rejected. This study found that Perceived Ease of Use and Perceived Usefulness are influenced by the constructs of the IS success model, except that the system quality variable on Perceived Usefulness is not significant. This study also found that TAM factors influence computer self-efficacy and satisfaction significantly. The anxiety variable is not significant to the TAM factor and the cognitive theory of Computer Self-Efficacy. The overall relationship between the variables analyzed has a small effect size.

Classification of Arrhythmia Electrocardiogram Signals Using Kernel Principal Component Analysis and Naive Bayes

Melinda Melinda — 2025-06-13

Arrhythmia is a cardiovascular disorder commonly detected through electrocardiogram (ECG) signal analysis. Classifying arrhythmias based on ECG signals remains challenging due to signal complexity and individual variability. This study aims to develop a more accurate and efficient method for arrhythmia classification. The proposed method utilizes Kernel Principal Component Analysis (KPCA) and the Naive Bayes algorithm for classifying arrhythmic ECG signals. KPCA is chosen because of its ability to reduce data dimensionality, which allows complex ECG signal processing and improves classification accuracy by minimizing noise. Naive Bayes algorithm is chosen because of its simplicity and computational speed, as well as its effective performance even with limited data. ECG signals are processed with KPCA to reduce data dimensionality and extract relevant features. The Naive Bayes algorithm is then applied to classify the ECG signals into four categories: Premature Atrial Contraction (PAC), Premature Ventricular Contraction (PVC), Left Bundle Branch Block (LBBB), and Right Bundle Branch Block (RBBB). Model performance evaluation employs metrics such as accuracy, sensitivity, specificity, precision, and F1-score. The Naive Bayes model achieves an overall accuracy of 97.67%, with the highest performance observed in the RB class at 99.33%. Additionally, the F1-scores for all classes range from 96.62% to 98.57%, demonstrating the model's capability to detect arrhythmias effectively. These results indicate that the combination of KPCA and Naive Bayes is effective for classifying arrhythmic ECG signals.

Analysis of Mental Health Disorders via Social Media Mining Using LSTM and Bi-LSTM

Binti Kholifah — 2025-06-13

Mental health disorders are a growing global concern, with many individuals lacking early detection and appropriate treatment. Mental illness can impact a person’s quality of life and often goes undetected until symptoms worsen. One contributing factor to this problem is the limited ways to detect mental disorders in their early stages. Social media, especially platform X, offers the potential to analyze users’ emotional expressions that may indicate a mental disorder, such as depression or anxiety. Psychological symptoms can be explored more broadly using Natural Languages Processing. This study optimizes several text pre-processing techniques to extract meaningful information from social media text. Then to convert words into number vectors, several word embedding methods are used such as Word2Vec, FastText, and Glove. Meanwhile, the classification process is carried out using LSTM and Bi-LSTM because they are considered capable of studying data sequence patterns such as sentence structure well. The results show that the addition of expanding contraction, emoticon handling, negation handling, repeated character handling, and spelling correction in the pre-processing text can improve model performance. In addition, Bi-LSTM with pre-trained FastText shows better results than other methods in all experiments with 86% accuracy, 87.5% precision, 84% recall, and 85.71% F1-Score.

Effectiveness of a Competitive Educational Game with a Game Controller in English Game-Based Language Learning

Galang Prihadi Mahardhika — 2025-06-13

Game-based language learning has emerged as a promising approach to language learning activities. Despite its potential, game-based language learning implementation concepts that emphasize player-to-player and player-to-game interactions have not been widely adopted. This study presents an educational game as a game-based language learning application that incorporates face-to-face interaction concepts and competitive game approaches to enhance player-to-player interaction. Additionally, the game utilizes a specially designed game controller to improve player-to-game interaction. The impact of the proposed educational game on the students' learning experience, gaming experience, and motivation was evaluated with a process conducted with 42 high school students (14 females and 28 males). The findings suggest that integrating concepts of face-to-face interaction in competitive game scenarios and the game controller design proposed in this study fosters social interactions among players, positively influencing students' learning experience, gaming experience, and motivation. Furthermore, the findings reveal that students prefer game controllers with microswitch buttons because they provide a physical feel that reduces errors during gameplay. This underscores the importance of ergonomic, easy-to-use game controller designs that minimize errors when playing educational games. By focusing on the interplay between player-to-player and player-to-game interactions, this study provides insight into designing interactive educational games that utilize interaction technology, particularly for language learning.

UI/UX Design for AR Card Game: Enhancing English Vocabulary Learning with Augmented Reality

Rio Krisdiawan — 2025-06-13

This study aims to develop and evaluate an Augmented Reality (AR)-based learning tool in the form of an AR Card Game to enhance English vocabulary acquisition among third-grade elementary school students, specifically on the topic of “Fruits and Vegetables.” The development process employed the User-Centered Design (UCD) methodology to ensure that the user interface and user experience (UI/UX) were aligned with the cognitive characteristics and needs of the target users. The prototype, designed using Figma, integrates interactive features including 3D object visualization, audio pronunciation guides, gamified elements, and physical card-based AR interaction. Evaluation was conducted through student questionnaires, teacher interviews, and classroom observations. The results indicate that the AR Card Game was positively received. A total of 85.07% of students reported improved understanding through 3D visuals, while 89.55% found the audio helpful in pronunciation. The gamification feature achieved a mean score of 4.18 (SD = 0.73), and a one-sample t-test revealed a statistically significant difference from the neutral score (p < 0.001), confirming its motivational impact. The coefficient of variation (17.48%) indicates consistent student responses. Teacher feedback also supported the tool’s effectiveness, although recommendations were made to improve navigation and enhance the evaluation component. Limitations of this study include its short-term implementation and focus on a single thematic domain. Future research is recommended to investigate long-term engagement, adaptive difficulty mechanisms, and the scalability of AR-based learning in broader curricular contexts. The findings underscore the potential of AR Card Games as effective and engaging tools for early language education in digital learning environments.

Cybersecurity Management Strategies for Smart Cities in Indonesia: Cultural Factors and Implementation Challenges

RG Guntur Alam — 2025-06-13

The implementation of smart cities in Indonesia presents significant cybersecurity challenges, particularly amid bureaucratic complexity, low digital literacy, and limited institutional capacity. This study explores cybersecurity management strategies in the context of Jakarta Smart City (JSC), emphasizing sociotechnical dynamics and embedded cultural-institutional factors. Employing a qualitative approach and the Actor-Network Theory (ANT) framework, this research examines four key moments in the stabilization of cybersecurity networks: problematization, interessement, enrolment, and mobilization. Empirical findings reveal that challenges such as fragmented governance, security awareness gaps, and limitations in technological adaptation are addressed through context-specific strategies. These include regulatory reforms, multi-stakeholder collaboration, hybrid governance models, and the localization of international standards, particularly ISO/IEC 27001. The study also incorporates Indonesia’s Personal Data Protection Law (Law No. 27/2022) as a foundational legal framework that supports the integration of regional cybersecurity policies. Rather than focusing solely on technical solutions, this research emphasizes the importance of aligning cybersecurity strategies with local norms, leadership structures, and user practices. The proposed strategic model contributes to the cybersecurity governance literature by integrating ANT perspectives with empirical insights from a developing country. It offers a locally adapted and scalable framework to guide policymakers and Smart city administrators in building resilient and culturally sensitive cybersecurity systems.

Analysis of Public Opinion on The Governor Candidate Debate Using LDA and IndoBERT

Ahmad Abdul Chamid — 2025-06-13

The gubernatorial candidate debate was broadcast live streaming through various YouTube channels, which attracted public attention. Many discussions and multiple conversations appeared in the comment’s column of each YouTube channel that broadcasted the debate. With the many public talks, it is undoubtedly interesting to analyze the contents of the conversation, as well as the expectations and input from the public. However, conversations in the form of text data will be challenging to analyze using conventional methods. So, in this study, public opinion will be analyzed using the topic identification and sentiment classification approaches. Topic identification is carried out to obtain accurate information about what the public is talking about, while sentiment classification is used to find out whether each comment contains positive or negative sentences. This research is novel because it uses data collected from various major media YouTube channels and qualitative analysis of the findings. This study uses public comment data taken from the KPU, NarasiTV, and KompasTV YouTube channels; the results obtained were 4.147 data. Preprocessing data carries out the process, identifying topics using the LDA method, evaluating the LDA model, then sentiment classification using IndoBERT and visualizing the results of the public opinion analysis. The results obtained were five topics with a perplexity value = -7.7909 and a coherence score = 0.5109. In addition, topic 4 is the most dominant compared to other topics, and there are 1.146 comments classified as positive sentiment and 504 negative comments. Topic 4 reflects how religion, culture, and frequently mentioned figures are perceived and discussed by the public, especially in relation to the gubernatorial election (pilgub) or gubernatorial candidate debates.

Land Price Distribution Prediction in Jakarta Using Support Vector Machine with Feature Expansion and Kriging Interpolation

Hadid Pilar Gautama — 2025-06-13

Fluctuations in land prices over time are very significant, especially in big cities, one of which is Jakarta. The increase in land prices is influenced by high demand, location-related needs, ease of access to various public facilities and crowds. Uncontrolled prices and lack of information about the distribution of land prices cause buyers to get land that is not in accordance with their needs. This study develops a land price distribution prediction system for Jakarta for 2025-2026 using Support Vector Machine (SVM) with time-based feature expansion and spatial interpolation. The SVM model with RBF kernel demonstrated superior performance, achieving 93.14% accuracy for 2025 predictions using the t-1 model. For 2026 predictions, the t-2 model achieved 83.33% accuracy. This approach involves utilizing one to two years of historical data and systematically selected features, ensuring more accurate and relevant predictions. Ordinary kriging interpolation visualizations revealed a significant shift in land price distribution patterns, indicating a decline in affordable land availability and an increase in high-value properties across Jakarta. The integration of SVM and kriging interpolation, coupled with comprehensive evaluation metrics, provides a robust methodological framework for predicting urban land price distributions. This system offers practical implications for informed decision-making in Jakarta's dynamic land market, enabling stakeholders to make efficient, budget-based property decisions. The research contributes significantly to urban planning by providing a comprehensive tool for understanding and predicting land price trends, which can assist various stakeholders in making informed property investment decisions.

Collaborative Filtering Modification Technology for Recommendation Systems in Smart Digital Agribusiness Marketplace

Setya Budi Arif Prabowo — 2025-06-13

The rapid transformation in the agribusiness sector, driven by globalization and digitalization, necessitates the adoption of intelligent systems to enhance performance, market accessibility, and decision-making processes. Despite the growing use of personalized recommender systems in e-commerce, geographical context remains insufficiently integrated into recommendation processes. This lack of geolocation awareness diminishes recommendation relevance and accuracy by overlooking geographical factors that influence user preferences. To address this limitation, this work aims to enhance the performance of recommendation systems in agricultural e-commerce by incorporating geolocation context through the integration of the Geo-Mod Neuro Collaborative Filtering (GMNCF) model into an Android-based application for agricultural products. The GMNCF model improves collaborative filtering by incorporating geographical region data to capture spatial user preferences and reduce data sparsity. Using Graph Neural Networks (GNNs), the model captures complex relationships among users, items, and geographic regions to generate more accurate recommendations. Experimental results reveal that GMNCF consistently delivers substantial performance improvements over baseline models such as NGCF, GC-MC, ASMG, and GCZRec. Compared to the strongest baselines, GMNCF demonstrates relative gains of approximately 4.9% in Precision, 5.9% in Recall, 5.6% in F1-Score, and 5.7% in Hit Rate. These improvements underscore the model’s effectiveness in capturing spatially influenced user preferences and strengthen the relevance of recommendations in the agribusiness e-commerce system. Furthermore, user testing with diverse respondents indicates high levels of satisfaction, particularly regarding location-based recommendation features and accessibility. These findings highlight the effectiveness of incorporating geographical region data into recommendation systems, which is particularly beneficial for geographically fragmented agribusiness markets.

Classification of Livin' by Mandiri Customer Satisfaction Using MLP with BM25 and TF-IDF Feature Weighting

Aina Mardiah — 2025-06-13

The increasing use of mobile banking applications such as Livin' by Mandiri requires analyzing customer satisfaction based on user reviews. This study classifies customer satisfaction level using Multi-Layer Perceptron (MLP) algorithm with two feature extraction methods, namely BM25 and TF-IDF. Data totaling 1,143 reviews were collected from Google Play Store and App Store. Three test scenarios were applied: (1) comparison of feature extraction methods, (2) application of Synthetic Minority Over-Sampling Technique (SMOTE), and (3) application of Synonym Replacement-based Easy Data Augmentation (EDA) technique. The evaluation results show that the combination of BM25 and data augmentation produces the highest performance with 97% accuracy and 98% precision, recall, and F1-score respectively. BM25 proved to be more effective in understanding the context of reviews, while data augmentation improved the quality of representation, especially on minority classes such as neutral sentiment. These findings make a real contribution to the improvement of Livin' by Mandiri digital services and serve as a reference for the development of review-based satisfaction classification systems in the digital banking sector.