International Journal of Advances in Data and Information System
Not a member yet
161 research outputs found
Sort by
Designing Enterprise Architecture for Vessel Services: A Case Study of PT Pelindo Jasa Maritim
Full text linkPT Pelindo Jasa Maritim (SPJM), as one of the subholdings of PT Pelabuhan Indonesia (Pelindo) providing marine and port supporting services, requires information technology to support service integration in line with the company’s strategic directions. Using the TOGAF framework, this study identifies five main business service clusters: Marine Services (Pilotage and Towage), Crewing, Port & Equipment Services, Dredging & Shipyard, and Utilities. Each cluster is designed to be supported by a single standardized core application that reflects the strategic direction of “one main application per cluster.” The proposed architecture combines a hybrid cloud approach with Internet of Things (IoT) technologies to enable advanced analytics and data-driven decision making. Data were collected through document analysis, interviews with key stakeholders, and observation of core business processes. This study adopts a Design Science Research (DSR) approach and evaluates the EA blueprint through expert-based validation against alignment, completeness, and implementation feasibility criteria. The study results in an enterprise architecture blueprint that defines the role of core applications for each service cluster and specifies the supporting technology platforms required to realize strategic directions. The design can serve as a reference for companies in the maritime industry.
Multi-Output Classification of Cognitive Levels and Topics in Indonesian Questions using Deep Learning and Transformers
Full text linkManaging large-scale digital question banks struggles with manual metadata labeling, especially when identifying material topics and cognitive levels based on the Revised Bloom\u27s Taxonomy. Current automated approaches usually treat these two attributes as separate tasks, which adds to the system\u27s complexity and computational load. This study introduces a multi-output classification method using a shared encoder architecture with two task-specific heads to predict topics and cognitive levels simultaneously. We performed experiments on 685 Indonesian junior high science questions, covering 15 topic labels and four cognitive levels (C1–C4), with an imbalanced distribution in which lower cognitive levels accounted for more than 75% of the dataset. To handle this imbalance, we applied Focal Loss to taxonomy classification, and class weighting was used in the comparison model. A comparative study involved CNN, BiLSTM, DistilBERT, and IndoBERT. Our results demonstrate that IndoBERT delivered the best performance, with F1-macro scores of 0.78 for topics and 0.71 for cognitive levels and showed better performance in minority classes compared to standard cross-entropy-based models. These findings suggest that an integrated multi-output approach can boost the efficiency and accuracy of question labeling and offers potential for integration into Computer-Based Test systems and e-assessment platforms in real time
Enhancing E-Insurance Mobile Applications through Cognitive Walkthrough and System Usability Scale (SUS): A Case Study of My Taspen Life
Full text linkThe digital transformation of the insurance sector has necessitated mobile services with superior usability. The My Taspen Life mobile application has experienced low user satisfaction attributed to its suboptimal user interface (UI) and user experience (UX). This study aimed to evaluate the application\u27s usability and propose a validated high-fidelity prototype employing a User-Centered Design (UCD) approach. A mixed-methods evaluation comprised a System Usability Scale (SUS) survey, a Cognitive Walkthrough with experts, and Usability Testing. The identified issues were mapped against Shneiderman’s Eight Golden Rules of Interface Design to develop an adequate solution. The initial results showed an average SUS score of 65.51, below the global SUS average, and identified critical constraints in layout, navigation, and visual representation. Following the redesign, the prototype was re-evaluated by 12 participants, equally distributed between employees and external users. The final evaluation demonstrated a significant improvement, achieving a SUS score of 83.54 (Grade A, "Excellent"). Task scenario testing further confirmed that the new interface significantly reduced user dissatisfaction. We conclude that implementing UCD-based intervention can address significant usability problems in e-insurance applications
Online Gambling in Indonesia: User Intention Factors Analysis
Full text linkThe rapid growth of online gambling posed a serious socio-economic threat in Indonesia, with transaction values reaching hundreds of trillions of rupiah. This study aimed to analyze the factors influencing the intention to engage in online gambling by integrating the Theory of Planned Behavior (TPB) with contextual external variables. Data were collected through a questionnaire from 187 Indonesian respondents with online gambling experience and analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM). The results indicated that online gambling intention was significantly and directly influenced only by perceived behavioral control and monetary gain. Furthermore, religiosity played a significant indirect role by strengthening perceived behavioral control, which in turn affected intention. The findings suggested that intervention strategies should extend beyond awareness campaigns to include targeted financial literacy education and the reinforcement of regulatory and technological barriers to disrupt the perceived ease and financial allure of gambling platforms
Comparasion Of Weather Classification Methods On Weather Images Using GLCM Features With Random Forest And Catboost Algoritms
Full text linkWeather image classification is an essential process for improving automated weather information systems. However, most existing studies rely on numerical meteorological data and rarely utilize the textural characteristics embedded in atmospheric imagery. This study addresses that limitation by applying the Gray Level Co-Occurrence Matrix (GLCM) for texture feature extraction combined with Random Forest (RF) and CatBoost algorithms for classification. The dataset, obtained from Kaggle, consists of 1,125 weather images categorized into four classes: cloudy, rain, shine, and sunrise. All images were uniformly normalized and augmented using four rotation angles (0°, 45°, 90°, 135°). GLCM features were extracted with a pixel distance of 1 and gray-level quantization of 8, generating four statistical attributes: contrast, correlation, energy, and homogeneity. Both algorithms were optimized through parameter tuning and evaluated using a 5-fold cross-validation scheme with an 80:20 split ratio. Results show that the Random Forest model (n_estimators = 100, max_depth = 10, random_state = 42) achieved the highest accuracy of 92.43% (±1.12), precision of 92.50%, recall of 92.43%, and F1-score of 92.42%. In comparison, CatBoost (iterations = 100, learning_rate = 0.1, depth = 6) achieved an accuracy of 68.88% (±2.31). The findings demonstrate that GLCM feature extraction combined with Random Forest offers superior stability and accuracy for weather image classification, providing a foundation for efficient and interpretable weather information systems
SARIMA-GARCH and LSTM Performance for Broiler Meat Price Forecasting: A Case Study in West Sumatra
Full text linkThe price of broiler chicken meat in West Sumatra is characterized by strong seasonality and high volatility. As a primary source of animal protein and a key contributor to regional inflation, accurate forecasting of these price fluctuations is essential for economic stability and policymaking. This study aims to compare the forecasting performance of the SARIMA-GARCH hybrid model against the Long Short-Term Memory (LSTM) model. The dataset consists of 1,198 daily observations spanning from 15 July 2022 to 24 October 2025, sourced from the National Food Agency (Badan Pangan Nasional). The results demonstrate that the SARIMA-GARCH model outperforms the LSTM model in terms of point forecast accuracy, as evidenced by lower prediction error metrics. Furthermore, the hybrid model successfully satisfies the statistical diagnostic criteria for volatility modeling by effectively resolving ARCH effects, ensuring the statistical validity of the residuals. While the LSTM model produces smoother long-term forecasts, the SARIMA-GARCH model effectively captures daily price fluctuations and indicates a modest upward trend over the next 28 days. These findings suggest that SARIMA-GARCH provides a more realistic depiction of short-term price movements for this specific regional market, offering a localized framework for stakeholders in West Sumatra to anticipate future market changes and maintain price stability
SPADE-LSTM: An Integrated Sequential Pattern Mining and Deep Learning for Badminton Next-Stroke Prediction
Full text linkBadminton rallies consist of complex and rapid stroke transitions that reflect players’ tactical decision-making. While prior studies have examined stroke patterns descriptively or applied standalone predictive models, limited research integrates interpretable sequential pattern mining with deep learning for next-stroke prediction. This study proposes an integrated SPADE–LSTM framework to analyze and predict badminton stroke sequences using a 10-class scheme (drive, dropshot, lob, netting, and smash for two athletes). Match data were transformed into structured stroke sequences and contextual features, then divided into training, validation, and test sets using a match–set–rally grouping strategy to prevent information leakage. Sequential patterns were first extracted using the Sequential Pattern Discovery using Equivalent Classes (SPADE) algorithm to capture frequent tactical transitions. These pattern-based features were subsequently used to train a Long Short-Term Memory (LSTM) model for multi-class classification. The proposed model achieved an accuracy of 88.68%, with weighted precision, recall, and F1-score of 0.9075, 0.8868, and 0.8851, respectively. Misclassifications were mainly observed in tactically similar stroke transitions and minority classes. The results indicate that integrating interpretable sequential pattern mining with deep learning provides both strong predictive performance and meaningful tactical insights for badminton performance analysis
Collaborative Governance in Smart City Makassar: Actors-Networks Across Capacity, Infrastructure, and Policy Domains
Full text linkThis study explores the collaborative governance approach in implementing Smart City Makassar, which faces challenges including low digital literacy, fragmented data, and weak institutional coordination addressing persistent issues of low digital literacy, fragmented data systems, and limited inter-agency coordination that remain underexamined in smart city governance research. Using a qualitative case study method supported by NVivo-assisted network content analysis, data was collected from policy documents, online news, and interviews. The research identifies actor roles, network patterns, and cross-domain integration of policies, infrastructure, and capacities to support smart city sustainability. Findings reveal that digital technologies operate as institutional infrastructures enabling cross-sector interoperability and data-driven coordinationin data-driven decision-making. A tripartite network structure emerges, with government as central orchestrators, private sector as co-innovators driving technological deployment, and citizens as active contributors shaping service responsiveness Matrix-query analysis indicates strong policy–infrastructure integration, highlighting regulation as the structural anchor of collaboration. The success of Smart City Makassar is shaped by the alignment of technical capacity development, regulatory coherence, and adaptive collaborative governance mechanisms. The study positions Makassar as a model for inclusive smart cities in Southeast Asia, contingent upon strengthened public trust and resilient digital infrastructure
Effectiveness Fine-Tuned Multilingual BERT Model for Sentiments Classification Toward Bali’s Cultural Attractions
Full text linkThis study examines the performance of a fine-tuned Multilingual BERT (mBERT) model for sentiment analysis of tourist reviews on Balinese cultural attractions. A multilingual dataset comprising 7,878 user-generated reviews from Google Maps and TripAdvisor was utilized to capture diverse linguistic expressions and visitor perspectives. The research methodology includes: (1) problem formulation and literature review; (2) dataset collection, preprocessing, and tokenization; (3) model training using mBERT as the baseline; (4) fine-tuning for domain adaptation; and (5) comparative evaluation with other Transformer models (XLM-Roberta and Distil-mBERT) and classical algorithms including Logistic Regression, Support Vector Machine, and Naïve Bayes. The results demonstrate a substantial improvement after fine-tuning. The baseline mBERT achieved 85.45% accuracy, while the fine-tuned model reached 92.13% accuracy with an AUC of 0.909, confirming the effectiveness of domain-specific adaptation. Although XLM-Roberta obtained slightly higher performance (93.15% accuracy, AUC 0.946), the fine-tuned mBERT showed stable and competitive results, making it the primary model of this study. Comparisons with classical methods further indicate that Transformer-based approaches provide more balanced and reliable sentiment classification. Sentiment distribution analysis reveals that tourist perceptions are predominantly positive, particularly regarding cultural authenticity and the quality of performances such as the Kecak and Fire Dance. Negative sentiments mainly relate to operational aspects, including crowd management, seating arrangements, and ticketing processes. Overall, this study provides empirical evidence that fine-tuned mBERT can effectively support data-driven evaluation of tourist experiences and deliver actionable insights for improving service quality and sustainability of Bali’s cultural touris
Development of A Control Delay Layer for Data Transmission Stability in Remote Patient Monitoring System
Full text linkRemote Patient Monitoring (RPM) systems generate continuous health data that must be reliably processed at the backend to support timely clinical decision-making. Many real-world RPM deployments rely on synchronous request handling, which can lead to service degradation and request failures under high concurrency. This study designed and evaluated a Control Delay Layer (CDL) as an application-layer mechanism to improve backend request-handling stability in a web-based RPM system. The proposed mechanism decouples data reception from permanent storage through temporary buffering and deferred batch processing while regulating data submission behavior at the service level. System behavior before and after CDL implementation was examined using controlled load testing under identical scenarios. The evaluation employed service-level performance metrics, including request failure rate, response time distribution, and computational resource utilization. Experimental results show that the baseline monolithic system experienced an average request failure rate of approximately 14% under peak load, whereas no request failures were observed after CDL implementation. The CDL enabled system maintained consistent response-time behavior and stable resource utilization at higher concurrency levels. These findings demonstrate that backend-level request-handling control can effectively enhance system stability under high load conditions without requiring device-level modifications, providing a complementary approach for scalable and resilient digital health systems