Publication Search

The proliferation of online gambling promotional comments on Indonesian social media has become a serious issue requiring fast and accurate automated handling. This study aims to implement a Hybrid Recurrent Neural Network (RNN) and Long Short-Term Memory (LSTM) method to classify online gambling comments and compare its performance with standalone RNN and LSTM models. The research utilized a dataset of 10,230 comments subjected to comprehensive preprocessing stages, including the normalization of non-standard language using a slang dictionary. Testing was conducted across three data-splitting scenarios: 90:10, 80:20, and 70:30. Experimental results demonstrate that the standalone LSTM model achieved the highest average accuracy of 97.45%. However, the Hybrid RNN–LSTM model showed significant superiority in terms of performance stability, yielding the lowest standard deviation (0.0027) and the smallest Coefficient of Variation (0.28%) across all scenarios. These findings indicate that while the LSTM architecture is highly effective at capturing short-text context, the Hybrid approach provides better robustness against fluctuations in data proportions, making it highly relevant for implementation as an automated detection system on social media.

Occupational Health and Safety (OHS) is a crucial aspect in manufacturing industries due to the high risk of workplace accidents caused by heavy machinery, chemical substances, and intensive production activities. Companies usually provide Standard Operating Procedures (SOP) as safety guidelines; however, SOPs are often delivered in long textual formats that are less engaging, making workers reluctant to read or difficult to understand quickly. This study aims to design an infographic-based SOP media as an effective visual communication tool to improve workers’ understanding of safety procedures. The research applies a qualitative method with a design approach through workplace observation, interviews with HSE personnel, literature review, and design validation using questionnaires. The results produce infographic media in the form of posters and signage presenting PPE usage procedures, hazard warnings, and evacuation steps using safety color codes, icons, and readable typography. The conclusion indicates that infographic SOP media is more effective than text-based SOP because it improves readability, comprehension, and workers’ memory of safety procedures.

This research benchmarks multiple machine learning (ML) algorithms for large-scale loan default prediction using a real-world dataset of 255,000 borrower records, where default cases represent only ~9–12% of total observations. The study addresses the persistent gap in comparative analyses of ML models that balance predictive accuracy, interpretability, and computational efficiency for credit risk assessment. Six algorithmic families were evaluated Logistic Regression, Random Forest, XGBoost, LightGBM, CatBoost, Artificial Neural Networks (ANN), and Stacked Ensemble—using standardized preprocessing, hybrid imbalance handling (SMOTE, class weighting, under-sampling), and comprehensive evaluation metrics (AUC, F1, Recall, Precision, PR-AUC, and Brier Score). Empirical results show Logistic Regression achieved the highest AUC of 0.732, outperforming nonlinear models under the baseline configuration, while LightGBM attained perfect recall (1.0) but low precision (0.116), indicating over-prediction of defaults. Gradient boosting models demonstrated robust calibration (Brier ≈ 0.114–0.116) and the best computational efficiency, with LightGBM showing the fastest training and lowest memory use. CatBoost exhibited strong recall but the slowest computation, and ANN underperformed on tabular data (AUC ≈ 0.56). The Stacked Ensemble delivered balanced results with AUC = 0.664 and improved overall stability. These findings confirm that boosting-based models, particularly LightGBM and CatBoost, offer superior scalability and calibration, whereas Logistic Regression remains a valuable interpretable baseline. The study concludes that effective default prediction requires integrating rebalancing, calibration, and threshold optimization to enhance recall and operational deployment reliability in large-scale credit ecosystems.

The increasing sedentary lifestyle in the digital era has the potential to cause various health problems due to lack of physical activity. One approach that can be taken to encourage physical activity is through the use of digital games with body movement-based control mechanisms. This study aims to develop a body gesture-based game character control system using a hybrid Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) model. CNN is used to extract spatial features from each video frame, while LSTM serves to model the temporal relationship between frames so that movement patterns can be recognized sequentially. The research method used refers to the Machine Learning Lifecycle stages, starting from data collection, preprocessing, model development, to implementation in the endless runner game genre. Testing results show that the CNN–LSTM model is capable of classifying body gestures and generating outputs that can be used as commands to control game characters. The implementation of this system enables more natural and interactive game interactions without conventional input devices, and has the potential to encourage players to lead a more active lifestyle.

International student enrollment is a critical driver of financial sustainability for Higher Education Institutions (HEIs). While advanced forecasting is standard in the corporate sector, its application in educational planning remains limited. This study addresses this gap by comparing the predictive performance of ARIMA, Facebook Prophet, and Long Short-Term Memory (LSTM) models. Using a publicly available annual dataset from a US-based institution (2000–2022), the analysis employed a strategic partition training on 2000–2017 and testing on 2018–2019 to validate models on stable, pre-pandemic data. Empirical results revealed that the statistical ARIMA (2,1,0) model demonstrated superior accuracy, achieving a Mean Absolute Percentage Error (MAPE) of 1.26%. Conversely, Prophet (11.81%) and LSTM (13.84%) struggled with the limited sample size, failing to generalize effectively compared to the linear approach. The findings suggest that for annual enrollment trends, parsimonious statistical models outperform complex deep learning architectures, providing administrators with a robust, accessible framework for data-driven strategic decision-making.

This study is motivated by the position of traditional music as a cultural element that not only functions aesthetically but also plays an important role in constructing social meaning, symbolism, and collective identity within a community. This article aims to examine the social meaning and function of Tagonggong music in the Tulude tradition of the Sangihe community in Londoun Village as a diaspora community. The research uses a qualitative method with a cultural ethnographic approach through participatory observation, unstructured interviews with traditional elders, Tagonggong players, and community leaders, as well as documentation. Data analysis was conducted descriptively and interpretatively using Alan P. Merriam's theoretical framework on the function of music. The results of the study show that Tagonggong music has a multidimensional function in the Tulude tradition, including symbolic communication, emotional expression, representation of cultural identity, reinforcement of social norms, and social integration and cohesion within the community.Tagonggong not only functions as an accompaniment to ceremonies, but also as a sacred musical instrument that connects the relationship between humans, ancestors, and God, while also marking the space and time of the ceremony. This finding confirms that Tagonggong is a living cultural archive that preserves the collective memory and ethnic identity of the Sangihe community in exile. Implicitly, this research enriches Indonesian ethnomusicology studies by positioning traditional music as an active social text in maintaining the cultural structure and solidarity of the communities that support it.

Postpartum depression (PPD) is a common and serious mental health disorder for mothers after givingbirth, which is a public mental health problem because it not only has a direct impact on the baby, but also on the family. The most common symptoms of PPD are overwhelming sadness, feelings of hopelessness and helplessness, moodiness, an inability to experience joy with the baby, serve anxienty, loss of appetite, poor concentration and memory, sleep disturbances, prolonged fatigue, and suicidal ideation (American Psychological Association, 2013). The design used in this study was quantitative by measuring the prevalence of postpartum depression in postpartum mothers using the EPDS form. The research was conducted at Gandapura Community Health Center, this location wa used as the research site because it had not been exposed at all regarding the use of the EPDS form as part of the initial assessment of midwifery services. A population is all elements that meet certain criteria for inclusion in a study. The population in this study was all postpartum mothers who gave birth in the obstetrics ward. The sample criteria in this study are divided into two, namely inclusion criteria and exclusion criteria. Conclusion: Postpartum guidance provided to mothers influences the risk of postpartum depression. Postpartum women who receive assistance are less likely to experience postpartum depression, and mothe who do not receive assistance are depression.

This study is motivated by the increasing need for accurate modeling and classification of one-dimensional signal data in intelligent systems. The rapid development of deep learning has led to the adoption of more adaptive and complex neural network architectures capable of capturing both temporal dependencies and local patterns in sequential data. This research aims to analyze and compare the performance of several deep learning models, namely Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and a hybrid Convolutional Neural Network–GRU (CNN–GRU) model for signal data classification. The research method employs a quantitative experimental approach involving data preprocessing, windowing, model training, and performance evaluation. Model performance is evaluated using accuracy, precision, recall, and F1-score metrics. The results indicate that the hybrid CNN–GRU model outperforms the other models, particularly in capturing local features and long-term temporal dependencies within signal data. These findings suggest that the integration of convolutional layers and recurrent mechanisms enhances feature representation and learning stability. This study is expected to contribute both theoretically and practically to the development of deep learning models for signal processing and time-series-based intelligent applications.

Stock price prediction remains a complex challenge due to the dynamic and non-linear nature of financial markets, especially for banking stocks like PT Bank Negara Indonesia (Persero) Tbk (BBNI). This study aims to optimize BBNI stock price forecasting by integrating an automated Extract, Transform, Load (ETL) pipeline with the Long Short-Term Memory (LSTM) algorithm within a data engineering framework. Historical data from 2019 to 2025 were processed through a structured ETL sequence—including data cleaning, feature engineering, and MinMaxScaler normalization—to ensure high data quality. The dataset was partitioned into 80% for model training and 20% for testing to ensure rigorous evaluation. The results demonstrate that the systematic ETL approach significantly enhances model stability and predictive accuracy compared to conventional methods. The LSTM model effectively captured long-term temporal dependencies, providing reliable trend forecasts with an impressive test accuracy, achieving a Root Mean Squared Error (RMSE) of 0.0354. This research underscores that integrating robust data engineering practices with deep learning is essential for building resilient financial decision-support systems.

This study examines the Ka’Bawakng dance as a cultural–religious expression of the Dayak Kanayatn community that has undergone a deepening of meaning through inculturation within Catholic liturgy. Originating in the Baliatn ritual, Ka’Bawakng dance initially functioned as a medium of healing and cosmological communication that affirmed the interconnectedness of human beings, nature, and Jubata. In its encounter with the Catholic faith, the dance was incorporated into the Eucharistic celebration as an offertory dance through symbolic reinterpretation grounded in the principles of inculturation articulated by the Second Vatican Council. This research employs a hermeneutical approach using participant observation, in-depth interviews, and documentary study, analyzed through thematic analysis. The findings indicate that Ka’Bawakng dance is not treated as a merely decorative liturgical element, but is understood as a language of prayer engaging the body, cultural memory, and the spirituality of the faithful. This integration deepens liturgical participation, strengthens the religious–cultural identity of the Dayak Kanayatn community, and creates a constructive space for dialogue between ancestral traditions and Christian faith. Nevertheless, sustaining the inculturation of Ka’Bawakng dance requires ongoing pastoral accompaniment to ensure fidelity to Church liturgical norms while respecting local cultural values.

Minister of Culture Fadli Zon's controversial statement, casting doubt on the existence of mass rape in the May 1998 Tragedy, sparked a strong public reaction, particularly from victims and human rights activists. This study aims to analyze Fadli Zon's statement as a form of identity politics in the context of contemporary Indonesian politics. This research uses a descriptive qualitative method with a discourse analysis approach to public statements, media coverage, and official documents such as reports from the National Commission on Violence Against Women and the Joint Fact-Finding Team (TGPF). The results show that Fadli Zon's statement is not merely a personal view, but rather a representation of identity politics that has the potential to obscure historical truth and reinforce social polarization. The identity politics that emerged in this discourse demonstrates how political power can influence the construction of national history and marginalize minority groups, particularly Chinese women victims of sexual violence. This study emphasizes the importance of historical honesty and the moral responsibility of political elites in preserving the nation's collective memory.

This research explores the impact of Cache Aware optimizations on signal processing pipelines in High Throughput computing systems. The growing demand for efficient memory management in modern computing systems, especially for data-intensive applications such as artificial intelligence (AI) and multimedia processing, necessitates the development of optimized memory hierarchies. Traditional memory systems often suffer from memory bottlenecks, significantly reducing the performance of these systems. This study investigates how memory hierarchy optimizations, particularly cache line aware optimization, dependency-aware caching, and adaptive cache replacement algorithms, can mitigate these challenges and improve system performance. Through analytical modeling and experimental benchmarking, this work evaluates various memory hierarchy configurations, including processing-in-memory (PIM) and three-dimensional integrated circuits (3D ICs), comparing them to conventional systems. The results demonstrate that Cache Aware optimizations lead to a reduction in memory access latency by up to 30%, while throughput improved by up to 40%. Additionally, cache hit rates increased by 25%, and energy consumption was reduced by up to 20%, highlighting the effectiveness of optimized memory management. The research contributes to the field by providing valuable insights into the design and implementation of efficient signal processing pipelines. It also identifies key challenges, including the need for dynamic occupancy mechanisms and DAG-aware scheduling algorithms, and suggests potential areas for future research, such as the exploration of collaborative caching approaches and further optimization of cache-adaptive algorithms. This work lays the foundation for more efficient, high-performance computing systems that can handle large datasets and complex tasks in real-time applications.

In high-performance computing (HPC) environments, the need to balance memory efficiency and query performance is crucial for ensuring optimal system performance. Traditional data structures, such as B-trees and hash tables, often prioritize either memory usage or query speed, leading to suboptimal performance in memory-constrained systems. This paper proposes a hybrid data structure that combines the strengths of multiple traditional data structures to optimize both memory usage and query processing speed. The proposed hybrid structure integrates cache-conscious algorithms, dynamic memory allocation, and compression techniques for intermediate query results. The approach is evaluated through extensive benchmarking tests comparing it to standard data structures like B-trees and hash tables under various workloads. Results show that the hybrid data structure reduces memory overhead by up to 30% while maintaining query processing speeds up to 1.5 times faster than conventional methods. Furthermore, the hybrid structure demonstrates robust performance across different types of queries, including both point and range queries, ensuring versatility and efficiency. The findings indicate that this hybrid approach provides a promising solution for HPC systems, where both memory efficiency and query speed are essential. Future research can explore extending the hybrid structure to distributed systems and emerging technologies, further improving its scalability and adaptability to new computational paradigms.

User reviews on the Shopee e-commerce platform represent an important source of information for understanding consumer perceptions of products and services. Sentiment analysis is commonly applied to classify user opinions into positive, neutral, and negative sentiment categories based on textual data. This study aims to analyze the performance of the Long Short-Term Memory (LSTM) method in sentiment classification of Shopee user reviews. The dataset used in this study consists of Indonesian-language user reviews that have undergone preprocessing stages, including case folding, text cleaning, tokenization, and stopword removal. The LSTM model was trained using preprocessed text represented as word sequences. Model performance was evaluated using overall accuracy and class-wise classification results. The experimental results indicate that the LSTM method achieved an overall accuracy of 87.62%. In addition, the classification performance for the positive sentiment class reached 95.27%, the neutral class achieved 4.96%, and the negative class reached 74.26%. These results demonstrate that the LSTM method performs well in classifying sentiment in Shopee user reviews, particularly for positive sentiment. This study is expected to provide insights and references for the application of deep learning methods in sentiment analysis of Indonesian e-commerce review data.

This study examines the teacher's strategy in guiding the memorization of prayer prayers in early childhood at RA Muslimat NU Banin Banat Manyar through a qualitative case study approach. The main focus is the application of the practice (repetition) and habituation method, which was observed for 6-8 weeks in 35 children in group AB (aged 4-6 years), including participant observation, in-depth interviews with 4 teachers and 5 parents, and analysis of RPP documentation and murojaah videos. The results show that the practice method is applied rhythmically daily (3x / day, 10-15 minutes), starting from simple pronunciations such as iftitah and ruku' with 20-30 repetitions per chain cycle, resulting in an average increase in memorization from 42% to 91%, with variations in singing and movements reducing boredom by 27%. Meanwhile, integrated habituation through congregational prayer routines (Dhuha, Zuhur simulation, Ashar), independent ablution, and home supervision, achieved 89% of children's independence in becoming mini imams after 21 days consistently, supported by verbal rewards and gender row rotation. The discussion confirmed alignment with Piaget's theory (preoperational stage) and Vygotsky's (ZPD scaffolding), where drills build sensory memory schemes while habituation forms permanent religious character ala Abdullah Nasih Ulwan. Supporting factors include parental collaboration and a conducive NU environment, overcoming the obstacle of low concentration. Practical implications recommend replicating this strategy in similar RAs to optimize the golden age of Islamic early childhood, with memorization retention of 8-10 basic prayer prayers.