In today's digital world, processing multilingual documents is critical for business, legal tasks, and information retrieval. This study describes a Multilingual Document Processing System that uses Optical Character Recognition (OCR) and Retrieval-Augmented Generation (RAG) to extract, query and summarize text in multiple languages. The system employs advanced OCR models to correctly recognize text from scanned documents, images, and handwriting in various scripts. By incorporating RAG, it improves comprehension and response generation, allowing users to retrieve and summarize information in English even when the original language is different. This approach takes advantage of recent advances in natural language processing, large language models (LLM), and multimodal AI to address challenges in multilingual data accessibility, knowledge synthesis, and real-time communication. The system provides a scalable AI-driven solution to improve document processing, eliminate language barriers, and increase global user engagement. AWS services support scalable document processing but cold starts in AWS Lambda hinder real time tasks.
Social media platforms have become the nerve center of campaign discussion in regards to global political landscape. Political candidates are wielding social media as an influential instrument to advance their own election campaigns. In this paper, we have perused the online exchanges and conversations among supporters of different political parties in the intense environment of US Presidential Election 2024. Sentiment scores, abusive speech and stance detection of the tweets have been considered for understanding the perspective of voters on social media platforms. Each post has been considered whether it is a hate speech which contributes in polarization in the various conversation obtained from mainstream social media platform X/Twitter. A novel model has been proposed that factors in sentiment, stance and and hate speech for opinion dynamics estimation. Our method has been evaluated on various publicly available datasets and satisfactory results have been obtained. It is observed that there is higher level of opinion polarization when hate speech is involved.
This study proposes a novel fixed-time tracking control methodology tailored for second-order multi-agent systems (MASs) operating over directed communication networks and experiencing communication delays. Recent studies have explored fixed-time consensus in MASs with nonlinear dynamics and communication delays. The control design employs a Lyapunov–Krasovskii (L-K) functional framework, integrating Hölder's inequality and a Lipschitz-like condition to ensure system stability. Notably, this approach eliminates the dependence on Linear Matrix Inequalities (LMIs), which are prevalent in traditional control strategies but often lead to increased computational complexity. By circumventing LMIs, the proposed method enhances computational efficiency, making it suitable for largescale MAS applications. The control protocol guarantees that all agents achieve consensus tracking of the leader's trajectory within a predetermined fixed time, regardless of initial conditions. Theoretical analyses are substantiated through numerical simulations, demonstrating the effectiveness and robustness of the proposed scheme in achieving rapid convergence and maintaining stability in the presence of communication delays.
This study developed the Digital Banks PH Notebook, a mobile application designed to support financial literacy among Filipinos by optimizing savings through high-yield digital banking platforms. The application featured a savings portfolio tracker, interest forecasting calculator, and savings goal management to address gaps in financial planning and savings behavior. Development followed a blended Agile methodology integrating Scrum, Extreme Programming, and Feature-Driven Development, ensuring iterative improvements aligned with user needs. Software quality was assessed using the ISO/IEC 25010 model, while qualitative feedback was analyzed through word cloud visualization to capture user sentiment and key focus areas. Findings indicated that the application effectively enhanced users' understanding of savings strategies and promoted responsible saving practices. The tool successfully connected the opportunities presented by digital banking with the practical requirements of financial education, providing users with actionable insights to manage their savings more strategically. By leveraging agile development practices and rigorous evaluation frameworks, the project demonstrated that technology-driven solutions can play a significant role in advancing financial literacy and supporting sustainable financial behaviors in an evolving digital economy.
This research presents the MySecureMap system, a geolocation-based recommender system designed for real-time air quality visualization and safetyfocused navigation. The system leverages the Air Quality Index (AQI) data provided by both government and private sources. The proposed mechanism has integrated the Long Short-Term Memory (LSTM) neural network's machine learning technique which is capable of predicting the AQI levels and providing the route recommendations. The ultimate goal of the proposed system aims to protect users from the hazardous air pollutants by automatically identifying the risk zones and suggesting the safe routes which are visualized on the map. This paper discusses the proposed approach and the system architecture, the design and the development of the proposed MySecureMap system, and the performance evaluation, highlighting its usability and effectiveness in enhancing the environmental awareness and public health. The performance evaluation results shown that the proposed system met the acceptable accuracy with Mean absolute percent error (MAPE) of 2.39% when comparing the predicted forecast value with the real AQI data observed for the next 12 hours.
An AI-driven academic path forecasting system is proposed to support data-informed advising and early academic intervention in higher education. In the Philippine context, where delayed graduation, student dropouts and lack of personalized academic guidance persist, machine learning in education offers a scalable and intelligent solution. The system combines three educational data mining techniques: a Long Short-Term Memory (LSTM) network for course sequence prediction, a decision tree classifier for student progress classification as regular or irregular and a K-Means clustering algorithm for grouping students based on academic trajectories. These models are developed in TensorFlow and deployed on a web platform built with CodeIgniter, enabling functionalities such as academic path forecasting, curriculum tracking and real-time risk alerts. Evaluation shows that the LSTM model achieves strong precision and recall in predicting next-term courses, while the decision tree classifier accurately detects off-track students with interpretable decision rules. K-Means clustering reveals meaningful groupings aligned with academic outcomes, further supporting early identification of at-risk learners. Confusion matrix analysis confirms high model accuracy across tasks. By integrating AI into higher education through course prediction, student classification and cluster-based insights, the system offers a practical framework for enhancing student success through targeted academic support.
Effective written communication is a crucial skill for students transitioning to industry, where accurate documentation of project activities and decisions are paramount. However, students' logbook entries are sometimes descriptive, unclear and incoherent. This might be due to them not having a critical reader in mind when writing. This paper evaluates the impact of a structured series of embedded workshops designed to enhance logbook writing skills among Applied Computing students to prepare them for their eightmonth internship under an Integrated Work Study Programme (IWSP). Generative AI (GenAI) took on the role of an interactive and critical reader: as a group of clients in a Requirements Engineering (RE) exercise (lesson 1), as a writing coach (lesson 2) and as an editor (lesson 3). Data was collected via a comparison of students' logbook entries in lesson three compared to lesson one and an end-of-workshop survey. The primary improvements observed between the students' logbook submissions were in critical analysis, content relevance and depth, and the integration of supporting ideas. The survey findings revealed that GenAI, when deployed as a user, helped generate ideas during brainstorming sessions and provided a basic and structured skeleton to kickstart the writing process. GenAI, as a coach, helped students be more objective by considering multiple perspectives. As an editor, it offered clarity and formatting suggestions, which helped students write more clearly and coherently. The findings underscore the effectiveness of AI-assisted workshops in preparing students for industry by developing essential written communication skills, thereby fostering better project documentation and reporting practices.
The carbon emissions from public utility vehicles (PUVs) in the Philippines are projected to contribute up to 80% of vehicle kilometers traveled and become a major source of emissions by 2035 without intervention. Recognizing the stochastic and condition-dependent nature of vehicular emissions, the research aims to identify and prioritize the factors influencing C O2 emissions using advanced analytical techniques. Emissions data were clustered using K-Means to identify distinct operational states, while Principal Component Analysis (PCA) reduced dimensionality and revealed key influencing factors. The Fuzzy Analytic Hierarchy Process (FAHP) was then applied to prioritize these factors, considering environmental, technical, and economic implications. Results showed a positive relationship between road slope and C O2 emissions $(r=0.3111)$ and an opposite relationship with respect to speed $(r=-0.3078)$, while acceleration had a minor positive effect (r=0.1330). FAHP assigned the highest weight to CO2 emissions (0.3589), followed by slope (0.3121) and speed (0.2972). Cluster analysis highlighted Cluster 0 as the most emission-intensive operational state, with an average C O2 level of 31394.25 g/km, moderate speed, and uphill road conditions. The integration of PCA and FAHP revealed that C O2 emissions and slope together accounted for over 67% of the emission profile importance. These insights inform the development of emission control strategies, eco-driving guidelines, and data-driven transport policies.
This study examines the flexural and thermal performance of fiber cement boards (FCBs) reinforced with recycled polyethylene terephthalate (PET) fibers as a sustainable alternative to conventional materials. Mixes with $5 \%, 10 \%, 15 {\%}$, and 20% PET content by weight were prepared and evaluated against a control. Flexural strength was measured after seven days using a three-point bending setup, while thermal behavior was assessed under simulated tropical conditions using a temperature-controlled environment. The 15% PET mix showed the highest flexural strength, although it remained below the commercial board benchmark, indicating the need for further optimization. The 5% PET mix demonstrated the most consistent thermal insulation, likely due to improved pore structure at lower fiber content. While PET fiber inclusion enhanced post-crack behavior and thermal resistance, early-age mechanical performance remained limited and did not meet industry thresholds. Nonetheless, incorporating recycled PET supports sustainable construction by reducing plastic waste and encouraging circular material utilization. Future research should include cost analysis, hybrid fiber development, and field-based durability evaluation.
Video augmentation is an effective strategy for improving the performance of action recognition models. A recent video augmentation strategy addresses scene bias by mixing human regions from one video with the background from another. However, this often produces artifacts due to limitations in the video mixing process, which degrade training quality. This study proposes a video augmentation strategy that produces compatible action-scene video pairs rather than choosing them randomly, to improve the quality of mixed videos. To achieve this, two compatibility metrics are introduced to guide this selection to significantly reduce the occurrence of visual artifacts and generate higher-quality augmented videos. Our method improves alignment between actions which leads to more effective augmentation. The performances are further enhanced by applying a temporal morphological operation to improve object detection consistency. Experimental results on the UCF101, HMDB51, and Kinetics-100 datasets show that our approach improves classification performance. Code is available at https://github.com/rendicahya/video-action-alignment.
Long-term ice dynamics of the Prince Harald system in Lützow-Holm Bay (LHB), East Antarctica, particularly before 1990, remain poorly investigated, hindering assessments of its historical contribution to regional mass balance. Here, we reconstructed high-resolution ice velocity fields (1973-1989) for its main components – Prince Harald 1, Prince Harald 2, and Prince Harald 3 – leveraging historical Landsat imagery with a robust photogrammetric method, a systematic framework that includes geometric correction, orthorectification, and hierarchical feature matching. Our integrated analysis, combining these velocities with assessments of basal melt, surface elevation change, ice front dynamics, and Passive Shelf Ice (PSI) behavior, reveals a stable state of the system. We find that this stability was characterized by consistent flow velocities, persistent surface thickening, minimal basal melt, and calving events predominantly within the PSI zone. These results indicate an overall trend of stability and net mass accumulation for the Prince Harald system during this period.
Low-power wide-area networks (LPWAN) have substantially improved the Internet of Things (IoT). LoRaWAN is a potential technology for IoT applications because it uses low-power, long-distance communication and offers excellent availability with low energy consumption. LoRaWAN power consumption can be reduced using the pure Aloha protocol at the MAC level. Optimizing orthogonal transmission parameters is still a major difficulty for enhancing network performance, even though they reduce packet loss and prevent collisions, especially in dynamic and heterogeneous networks. However, the challenge of random channel selection in LoRaWAN communication often leads to inefficient resource utilization and degraded network performance. This paper proposes a novel game-theoretic approach for optimal channel selection in LoRaWAN networks. Our method leverages real-time Received Signal Strength Indicator (RSSI) data and a non-cooperative game theory model to dynamically select channels, thereby improving throughput and reducing packet loss. Through extensive simulations and a realworld testbed, we demonstrate that our proposed mechanism outperforms existing approaches such as the Online Decision algorithm and MFMSF. Specifically, it achieves up to $22 \backslash \%$ improvement in throughput, $15 \backslash \%$ higher packet delivery ratio, and $18 \backslash \%$ reduction in latency, while consuming up to $25 \backslash \%$ less energy under heavy and dynamic traffic conditions. This work offers a significant advancement in enhancing the scalability and reliability of LoRaWAN networks, paving the way for more efficient IoT communications.
Small-signal modeling of Gallium Nitride (GaN)-based High Electron Mobility Transistors (HEMTs) requires accurate self-heating and charge trapping representation. A small-signal model is proposed which simplifies yet accurately represents drain-lag behavior, improves the extraction flow and simulation precision of the physics-based compact model for the 250 nm device. To incorporate the thermal effects in the proposed small-signal model, the conventional ASM-HEMT model is referred and modified suitable. The model includes R ${}_{\text{sub }}=20 \Omega$ and C ${}_{\text{sub }}=100 \text{fF}$, along with a thermal network defined by $\mathbf{R}_{\mathbf{t h}}$ and $\mathbf{C}_{\mathbf{t h}}$ to capture transient thermal behavior. Pulsed I-V measurements and temperature-dependent characterization are used to validate the proposed model. Pulsed I-V measurements show peak Ids values of 230.45 mA at $\mathrm{V}_{\text{dsq}}= 8 \mathrm{V}, 228.15 \text{mA}$ at 15 V, and 213.44 mA at 28 V as compared to the conventional ASM-HEMT reference model. The investigation is carried out in a calibrated Cadence Virtuoso simulation environment.
Perovskite solar cells (PSCs) face significant challenges such as instability, high recombination, poor charge transport, and mismatched band alignment, which limits their power conversion efficiency (PCE). These issues are more pronounced in multi-junction or heterojunction configurations. This study addresses them by proposing an advanced and novel dual-absorber thin-film heterojunction structure (THSC) is Au/CBTS/BiFeO3/CIGS/PDINO/FTO2/Ni2 using the Key Materials such as CBTS and BiFeO3 improves high carrier mobility, optimal layer thickness, controlled defect levels, tailored doping concentrations, and minimized interface defects, CIGS enhances light absorption and provides efficient charge carrier generation due to its tunable band gap. This contributes to reduced recombination losses, improved photocurrent, and overall device efficiency. The structure of the THSC device was optimized and simulated to have remarkable photovoltaic parameters, impressive performance, with a PCE of 40.10%, Jsc of 35.81 mA/cm2, Voc of 1.31V and FF of 88.51%. The novel architecture proposed guidance for future experiments and simulations aimed at developing high-efficiency, stable thin-film heterojunction solar cells in next-generation photovoltaics.
The researchers introduced an interactive projection platform that enables users to control on-screen content through natural hand gestures, enhancing the way people interact with digital media. This innovative system integrates gesture recognition technology, high-definition cameras, and projectors to detect user movements in real time, allowing seamless navigation of presentations and educational materials. By eliminating the need for traditional remote controls and whiteboards, it fosters a more engaging and dynamic experience in both academic and professional environments. The system's core advantage lies in its precise gesture recognition and smooth performance, powered by advanced machine learning algorithms based on Google MediaPipe and supported by high-quality hardware components. Its modular design allows for quick setup, instant response times, and handheld functionality, ensuring adaptability across various settings. Cost-effectiveness was a key consideration, with extensive testing and refinement addressing detection accuracy, system compatibility, and environmental challenges. Ultimately, this gesture-based interface aims to transform presentation methods by promoting immersive, intuitive, and highly interactive user experiences.
In the Philippines, traditional farming methods are still used. Modern techniques like drip irrigation, controlled-release fertilizers, and digital tools are helping farmers improve efficiency, reduce waste, and increase profits. This research focuses on developing a device to automate crop cultivation and harvesting, offering an alternative to traditional practices and addressing the need for innovative solutions in agriculture like the Philippines' farming context. The system uses an LCD to show the soil moisture levels in real time, two servo motors for automated soil cultivation, and a micro servo for regulated seed dispensing. Furthermore, an ultrasonic sensor tracks the height and development of plants, offering useful information for improving crop management. All servos are synchronized by an infrared (IR) controller, which guarantees precise and efficient planting task execution. The project was evaluated by experts and attain the score of 3.88 which is described as Excellent. This means that the project is a flexible and scalable solution for to decrease human labor, improve planting accuracy, and promote sustainable agricultural methods.
Internet of Things (IoT) and Blockchain integration are having a huge impact on the future of technological progress. IoT has progressed from an emerging notion to a widely used technology, shaping the future of digital connectivity. With billions of networked IoT devices generating large amounts of data, efficient data management becomes critical. Blockchain has been explored extensively to enhance security in IoT networks however, its scalability limitations become evident when handling large-scale deployments. Sharding is recognized as a promising approach to improve blockchain scalability by partitioning the network into multiple independent groups. These groups, called shards, process transactions in parallel, increasing throughput while reducing communication, computation, and storage over-head. Despite its advantages, many existing blockchain sharding models rely on static algorithm, which fail to accommodate the dynamic nature of blockchain networks. Factors such as variable node involvement and possible security concerns present problems that static sharding cannot solve. To address these restrictions, deep learning provides a strong solution for dynamic and multidimensional sharding in blockchain-based IoT systems. Deep learning, with its capacity to understand complex patterns and adapt to changing network circumstances, can improve the efficiency, security, and scalability of blockchain-powered IoT networks. This article proposes a deep reinforcement learning-based dynamic shards in blockchain IoT applications to overcome scalability difficulties.
The growth of the renewable energy target has increased the need for a powerful solar system. However, those caused by solar power production itself from irradiation, temperature, and the state of the solar panels present tough challenges in prediction, beneficial use and optimization. This study aims to accurately model solar power generation forecasting using machine learning models like Linear Regression, Decision Tree Regressor and Random Forest Regressor. The study investigates the influence of the environment on power output, underlining the daily diurnal variability of irradiation and temperature. Here, we describe how time-series modelling and clustering algorithms help identify patterns of solar power behavior distilled from data collected across multiple months and locations. The results show that compared to other models, the Random Forest model boosted forecasting accuracy by capturing non-linear overheads and minimizing the variance. The best-case scenario yields 92.5 W peak DC power generation, and the worstcase scenario shows 12 W minimum peak DC power generation. Moreover, solar power generation is heavily affected by temperature and irradiation changes, as in daily temperature fluctuations from 2 0° C to 8 5° C, in addition to combining anomaly detection using Z-Score analysis and Isolation Forest algorithms to identify problems such as degradation of panels and fault of sensors. These findings emphasize the necessity for a near realtime model and complex forecasting models to mitigate issues associated with renewable energy precursors to increase solar energy productivity and efficiency.
This paper outlines an FPGA-accelerated perception sub-system for real-time object detection employing the You Only Look Once (YOLO) v5 model. The system uses the Deep Learning Processing Unit (DPU) B4096 implemented on the Xilinx Kria KV260 platform for efficient deep learning inference. The implementation shows a $7.5 \times$ speedup over traditional CPU solutions, achieving up to 59.94 Frames Per Second (FPS) for the YOLOv5 Nano model and 17.4 FPS for the YOLOv5 Large model. The system further shows high detection accuracy with a mean Average Precision (mAP) of 0.76 and an average Intersection over Union (IoU) of 0.72 for the intelligent traffic perception system. The implementation incorporates optimized data pipelines, quantized models, and high-throughput inference methodologies within the Kria Docker environment, utilizing OpenCV, Xilinx Runtime(XRT), and Vitis AI Runtime (VART). The results highlight that an effective hardware-software co-design approach, combining high throughput, optimal resource utilization, and reliable inference accuracy, significantly enhances the performance of FPGA-based AI (Artificial intelligence) workloads in applications such as surveillance, industrial automation, and production line safety.
The accuracy and reliability of the blockchain networks simulation depend on the realistic network and simulator configuration. However, many existing blockchain simulators may use outdated parameters that fail to reflect the dynamic and constantly evolving nature of real-world blockchain systems. Hence, a simulator validation is crucial before starting the process of improving blockchain performance. In this paper, we collect the recent blockchain-related data from different sources and re-parametrize the blockchain parameters. A different network setting, namely a 7 regions simulation, was proposed to be used in the blockchain simulator to improve the reliability of the blockchain simulator. We then compare the block propagation time between the proposed settings with the original settings and the results of the DSN network, which reflects the Bitcoin network using the rate of change rule. The results show that the proposed settings are slightly closer to the Bitcoin network compared to the original settings in terms of block propagation time. The results are then validated using the standard deviation to verify consistency. We also suggest further collecting the data for a longer duration to achieve higher reliability on the simulator and discuss the future works.
