Journal of Soft Computing Exploration

Topic modelling analysis of public policy narratives on prabowo-gibran in national news

Lukmanul Hakim, Anggi Yudistira Aditya, Muharman Lubis — Wed, 31 Dec 2025 00:00:00 +0000

The rapid acceleration of digital transition has become an inevitable reality of the modern era. The proliferation of online communication platforms, news portals, and heterogeneous data formats has substantially increased big data volumes, leading to large-scale collections of unstructured data. This study aims to analyze dominant public policy–related topics concerning the Prabowo–Gibran administration by applying topic modeling techniques to national online news media. Latent Dirichlet Allocation (LDA) and Non-negative Matrix Factorization (NMF) were employed as unsupervised learning approaches to extract latent semantic structure from a corpus of 200 credible news articles collected through URL fetching using Python 3. Data preprocessing included text cleaning, tokenization, bigram and trigram construction, and the development of a dictionary and corpus. Model performance was evaluated using topic coherence metrics, yielding scores of 0.3709 for LDA and 0.68 for NMF. To examine temporal dynamics, the dataset was divided based on the official inauguration date of the president and vice president, enabling a comparative analysis of dominant topics before and after the inauguration. Topic similarity across both periods was measured using cosine similarity, with the highest similarity score of 0.663 observed between Topic 4 in the pre-inauguration period and Topic 1 in the post-inauguration period. The findings provide insights into evolving media discourse and policy-related topic trends across the two periods, demonstrating the potentials of topic modeling in analyzing large-scale unstructured news data for diverse purposes to bridge computational science and empirical evidence of social science.

Web based IoT monitoring system for ultrasonic water flow measurement using ESP32-S3 and cloud database

Mon, 29 Dec 2025 00:00:00 +0000

Efficient water management is crucial for ensuring sustainable resource utilization and reducing water losses in both industrial and domestic applications. This study presents the design and implementation of a smart water monitoring system based on an ultrasonic flow meter, which enables accurate, real-time measurement of water flow without physical contact with the medium. The proposed system integrates ultrasonic sensors with a microcontroller-based data acquisition unit and wireless communication to transmit flow rate, volume, and consumption data to a cloud-based monitoring platform. The system was tested in various flow conditions to evaluate accuracy, stability, and energy efficiency. Experimental results demonstrate that the ultrasonic flow meter achieved a measurement accuracy of ±1% compared to a reference turbine flow meter, while maintaining minimal power consumption. Furthermore, the integration of Internet of Things (IoT) capabilities allows remote monitoring, anomaly detection, and data logging for long-term analysis. The results indicate that this ultrasonic-based monitoring system provides a reliable and non-invasive solution for smart water management, offering potential applications in household metering, agricultural irrigation, and industrial fluid monitoring.

Application of the TAM model for assesing the acceptance of IoT technology in a residential security application

Rajib Ghaniy, Binanda Wicaksana, Fahmi Arnes, Laras Melati, Helena Septiana — Fri, 09 Jan 2026 00:00:00 +0000

Residential crime continues to be a significant concern, and traditional security systems relying oHousing is an area vulnerable to crime, especially if it is not supported by an adequate security system. Many housing complexes still rely on conventional security systems that only involve officers without technological support. Therefore, the application of technology, especially the Internet of Things (IoT), is needed to improve housing security systems. The success of a system is largely determined by the level of user acceptance, which can be measured using the Technology Acceptance Model (TAM). This study aims to measure user acceptance of an IoT-based housing security system using the TAM model. Data were obtained from 100 respondents and analyzed using the PLS-SEM method to test the research hypotheses. The results showed that four hypotheses had a significant relationship, namely the relationship between Subjective Norm (SN) and Perceived Usefulness (POU), Perceived Ease of Use (PEU) and POU, PEU and Attitude Toward Use (ATU), and POU and Behavioral Intention (BEI). Meanwhile, the other four hypotheses did not show a significant relationship.n manual monitoring are often insufficient in addressing modern security challenges. With the rapid development of Internet of Things (IoT) technology, digital security solutions offer new opportunities for improving surveillance and access control within housing environments. This study aims to assess user acceptance of an IoT-based residential security application by applying the Technology Acceptance Model (TAM). A quantitative survey method was used, involving 100 respondents who evaluated the prototype after testing it directly. Data were analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM). The results indicate that Perceived Ease of Use significantly affects both Perceived Usefulness and Attitude Toward Using, while Perceived Usefulness strongly influences Behavioral Intention. However, Attitude Toward Using shows a marginal impact on Behavioral Intention, and Behavioral Intention does not significantly predict Actual Use. These findings reveal the dominant factors influencing acceptance and highlight areas for improvement in IoT-based security applications.

Comparative performance analysis of YOLOv8 small and larger for real-time website-based monitoring

Mon, 02 Feb 2026 00:00:00 +0000

River trash pollution in Indonesia demands efficient real-time monitoring solutions. By using deep learning, while YOLOv8 is a promising model for implementation, it is a family of models with each variant differ in accuracy, speed, and computational demand. Among these, YOLOv8s and YOLOv8l come out as potential variants due to the balance between speed, accuracy, and computational demand. Therefore, to address this gap, this study intends to compare the YOLOv8 small (YOLOv8s) and YOLOv8 large (YOLOv8l) variants for real-time, website-based river trash monitoring systems, aiming to identify the optimal balance between accuracy and inference speed for practical real-time deployment. A combined dataset of 66 images, consisting of 86% images from Kaggle’s 2024 dataset and 14% AI-Generated images from Krea AI, was augmented using Roboflow to produce 591 annotated images. Both models were treated equally with the same dataset and method. The evaluation was conducted using Precision, Recall, mAP50, mAP50-95, Inference Speed, and Frame per Second (FPS) speed. Consequently, YOLOv8s exceeded YOLOv8l, achieving approximately 20% higher precision, 30% better detection quality, and 20% higher mAP across IoU threshold, and 150%-340% faster FPS. This finding indicates that YOLOv8s offer better accuracy and speed trade-off for real-time implementation than YOLOv8l. Moreover, successful integration to the website with real-time stream and threshold alert, confirm its feasibility for proactive waste and flood management.

Literature analysis on the role of artificial intelligence in addressing fraud in digital financial services

Ayu Faza Islami, Magdalena A. Ineke Palereng — Mon, 09 Feb 2026 00:00:00 +0000

The rapid growth of digital financial services has significantly increased fraud risks, threatening the security of global financial systems. This study addresses the limitations of traditional fraud detection by analyzing the role of Artificial Intelligence (AI) as a real-time prevention mechanism. Using a Systematic Literature Review (SLR) of 24 scientific articles published between 2019 and 2025, this research evaluates AI’s effectiveness, implementation challenges, and its synergy with Big Data, Blockchain, and AutoML. The findings demonstrate that AI models, particularly Deep Learning and Machine Learning algorithms, provide superior accuracy in anomaly detection compared to conventional rule-based systems. However, implementation is often hindered by data scarcity, high false-positive rates, and infrastructure costs. The study concludes that a collaborative framework—integrating AI for predictive analysis, Blockchain for data integrity, and Big Data for scalable processing—creates a more robust and adaptive defense against sophisticated financial crimes. These insights provide a conceptual foundation for developing more comprehensive digital security ecosystems.

Comparative study of pre-trained RoBERTa sentiment models and zero-shot LLM on indonesian and english texts

Akmal Faiz Agiputra, Jumanto Unjung, Budi Prasetiyo, Nurrizky Arum Jatmiko — Mon, 09 Mar 2026 00:00:00 +0000

The growth of user-generated content on social media has increased the need for effective sentiment analysis methods. Although fine-tuned transformer-based models and zero-shot large language models (LLMs) have both been applied to sentiment classification, comparisons across languages under unified evaluation settings remain limited. This study examines the trade-offs between task-specific fine-tuning and instruction-based zero-shot inference for multilingual sentiment classification. Experiments were conducted using two publicly available Twitter sentiment datasets in Indonesian and English, each annotated into three sentiment classes. Fine-tuned RoBERTa-based models were evaluated on full test sets, while all models, including a zero-shot LLM, were compared using an identical controlled subset. Performance was assessed using accuracy and macro-averaged precision, recall, and F1-score, with macro F1-score as the primary metric. The results show that fine-tuned RoBERTa-based models achieve stable and balanced performance across sentiment classes, with monolingual models consistently outperforming multilingual variants. Under controlled evaluation, zero-shot LLMs demonstrate competitive performance in English but remain less effective in Indonesian, indicating that their effectiveness is influenced by language resource availability. Overall, this study provides a controlled comparison of the strengths and limitations of fine-tuned and zero-shot approaches for multilingual sentiment classification.