Text-to-Structured Query Language (SQL) enables natural language interfaces to databases, yet its performance in low-resource languages such as Vietnamese is limited due to scarce annotated data and translation noise. Existing few-shot prompting methods for large language models (LLMs) typically rely on semantic similarity or random sampling, which often results in redundant or syntactically misaligned in-context examples. To address this limitation, we propose Vietnamese Intelligent Retrieval and Re-ranking (ViR²), a two-stage, unsupervised example selection framework that first performs semantic retrieval and then applies beam-guided re-ranking to jointly optimise syntactic alignment and intra-set diversity. ViR² is fully language-agnostic and schema-free, making it suitable for low-resource and cross-lingual Text-to-SQL scenarios. Experiments on ViText2SQL demonstrate that ViR² achieves 23.33% Exact Match (EM) and 87.83% component-level F1, outperforming Random, Skill-KNN, ASTRES, and DICL by up to 5 EM points. Ablation studies confirm that Part of Speech (POS) -based syntactic re-ranking and beam search are the principal contributors to performance gains, while maintaining practical single query latency. By combining semantic relevance with syntactic diversity optimisation, ViR² provides a robust and extensible approach for few-shot Text-to-SQL in low-resource and multilingual environments.
Austroasiatic, a major ethnolinguistic family in Vietnam, accounts for 90% of the census population size. Apart from the majority Kinh group, 24 other Austroasiatic minorities account for 4.67% of the population, many of which remain understudied. Here, the genetic characteristics of three Vietnamese Austroasiatic ethnic groups (Bana, Hre, and Mnong) were investigated by analysing the complete mitochondrial DNA (mtDNA) of 119 individuals. A total of 6,361 polymorphic sites were identified, with Bana exhibiting the highest molecular diversity (53.64±3.51 variants per individual). Compared to hypervariable segments (HVS), complete mtDNA sequences enhanced phylogenetic resolution and accuracy. The maternal haplogroup profile consisted of 40 lineages, with 32 of the detected lineages being population specific, six shared between two populations, and two shared among all populations. The macro-haplogroup components included B, C, F, M, N, R, and Z, with 66-80% of each population assigned to haplogroups B and M. The most prevalent haplogroups were M20 (14.29%), B5a1a (6.72%), and M74b1 (6.72%). While lineages of haplogroup F were relatively more frequent in Bana and Hre, lineages of N were exclusive to Mnong, highlighting haplogroup distinctions within the Vietnamese Austroasiatic family of the Central Highlands. These findings contribute to reconstructing the complex history of migration and settlement in Vietnam.
This article presents the process of developing an innovative skincare product derived from natural ingredients, specifically Aloe vera and perilla leaf essential oil (Perilla frutescens). The product is formulated in a gel-based form, making it lightweight, non-greasy, and suitable for a wide range of skin types, including normal, oily, combination, and especially sensitive skin. By combining clean, affordable, and easily accessible ingredients, the research has successfully created a high-quality skincare solution with excellent moisturising, protective, and regenerative properties at a relatively low production cost. The gel is rapidly absorbed into the skin, providing immediate hydration while delivering essential nutrients that help maintain healthy and balanced skin. Aloe vera extract is well known for its soothing, anti-inflammatory, and healing effects, whereas perilla leaf essential oil contributes antioxidant and skin-brightening benefits. Together, these ingredients help reduce irritation, improve skin texture, and support natural skin recovery. In addition, the product incorporates nanoparticles such as gold (Au) and titanium (Ti), which enhance skin regeneration, reduce dark spots, stimulate collagen production, and promote brighter, smoother, and healthier-looking skin with long-term use.
Panax vietnamensis var. fuscidiscus, commonly known as Lai Chau ginseng, is a high-value medicinal plant native to the mountainous regions of northern Vietnam. To develop effective management strategies for Lai Chau ginseng, this research aimed to comprehensively assess the pests and diseases affecting its cultivation. In 2024, a survey was conducted in Sin Ho, Lai Chau province. A total of 48 symptomatic plant samples were collected and evaluated based on morphological characteristics and visible symptoms. Four major symptom groups were identified: insect-induced leaf damage (32%), damping off (29%), leaf spots (20%), and root rot and wilting. Two key insect pests were identified: Chinavia hilaris (green stink bug) and Pseudococcus citri (mealybug), which caused significant damage to flowers, fruits, and petioles. Additionally, five pathogens were identified, including Cladosporium paeoniae (leaf spot), Puccinia sp. (rust), and three soil-borne fungi (Fusarium solani, Pythium ultimum, and Rhizoctonia solani), responsible for root rot and damping-off. These findings provide important baseline data for developing Integrated Pest Management (IPM) strategies to support the sustainable cultivation of Lai Chau ginseng.
Metro line No. 1 is the first urban railway line in Ho Chi Minh city, traversing densely populated urban areas and major passenger attraction centres. While it offers the potential for modern transport connectivity, operational experience has revealed that the existing transport infrastructure in adjacent areas does not fully meet passenger travel demand and has not kept pace with the city’s rapid urban development. This study focuses on assessing the scientific basis and current state of transport organisation, with the aim of applying the Transit-Oriented Development (TOD) model to Metro line No. 1. The research develops and compares two simulation scenarios: (i) the current situation scenario, reflecting existing conditions of transport connectivity, and (ii) the TOD scenario, applying TOD-based transport organisation solutions. Transport network performance indicators analysed from the simulation results include travel time, level of service, accessibility, and traffic volume. Based on the analysis of current conditions, the study proposes solutions and evaluates their effectiveness through simulation results to formulate a set of TOD-based transport organisation measures. These proposals are expected to enhance the operational efficiency of Metro line No. 1 while promoting the sustainable development of Ho Chi Minh city’s urban areas.
The judicious selection of laser cladding (LC) process parameters is essential for the clad bead geometry. In this research, the Taguchi-Grey relational method was chosen to enhance the quality of single-track bead geometry and to perform multi-objective optimisation of process parameters. The Taguchi method was selected to construct an L16 orthogonal experiment. The impact of significant LC parameters (laser power (P), scanning speed (S), and powder feeding rate (F)) on the responses (width (W), height (H), penetration (Pene), and dilution rate) was analysed based on the analysis of variance (ANOVA). The findings indicated that the scanning speed was the most important factor affecting the ratio of layer W, H, Pene, and dilution rate of the cladding. Subsequently, grey relational analysis (GRA) was employed to consolidate the four response variables into a singular grey relational grade (GRG) for quantification and optimisation purposes. Finally, the optimum cladding process parameters were obtained. The face-centred cubic (FCC) single-phase coating was observed using X-ray diffraction (XRD), scanning electron microscopy (SEM), and energy dispersive spectroscopy (EDS) to analyse the microstructure and chemical composition of the coating. Moreover, hardness measurements exhibited a gradual increase from the substrate (197 HV) to the cladding layer (224 HV), attributed to the distribution of Cr and Mo, solid solution strengthening, and microstructural refinement, demonstrating improved mechanical performance.
The study proposes a novel Meshfree Method of Lines (MFMOL) for the numerical solution of three-dimensional advection-diffusion equations (ADEs) used to model solute transport in porous media. The method involves discretising the spatial variables of the model equations using an Augmented Radial Basis Point Interpolation Method (ARPIM), while the temporal variable remains continuous. This results in a system of ordinary differential equations (ODEs), which is numerically integrated using the MATLAB ODE solver. In solving solute transport in porous media, traditional mesh-based methods such as the Finite Element Method (FEM), Finite Difference Method (FDM), and Finite Volume Method (FVM) often encounter challenges such as numerical instabilities and distorted meshes caused by complex geometries and the presence of convective terms. To ensure accuracy, various stabilisation techniques are required, which increase computational effort, cost, and setup time. To overcome these challenges, the new MFMOL approach was proposed and applied in strong-form formulations without stabilisation techniques to solve 3D advection-diffusion problems in homogeneous and isotropic porous media. The results obtained were in good agreement with existing exact solutions, establishing the accuracy and efficiency of the new method and demonstrating its advantages over traditional mesh-based methods for solving problems involving complex geometries.
Vietnam’s Nationally Determined Contribution (NDC) sets a target of reducing total greenhouse gas emissions in the transportation sector by 34.33%, equivalent to approximately 15.66 million tonnes of CO₂, by 2030. This study was conducted to forecast the potential for achieving that target through the application of a fuel consumption limit for cars with fewer than nine seats and for motorbikes in Vietnam. The results show that the number of new cars increased rapidly from 159,289 vehicles in 2016 to 347,287 vehicles in 2022. Similarly, the number of new motorbikes increased from 2,917,086 units in 2020 to 3,202,554 units in 2022. The average fuel consumption of vehicles with fewer than nine seats and an average engine displacement of 1,748 cc was 6.98 L/100 km during the 2016-2022 period, while the average fuel consumption of motorbikes was 1.96 L/100 km. By applying an annual reduction rate in fuel consumption of approximately 4%, based on current levels, and under GDP growth scenarios ranging from 8 to 11% (forecast for 2025-2030), the total reduction in CO₂ emissions is projected to range from 15.73 to 16.53 million tonnes. This reduction would meet the target set in Vietnam’s updated 2022 NDC.
Acetylcholinesterase (AChE) inhibitors constitute a well-established class of therapeutic agents that provide symptomatic relief in Alzheimer’s disease, underscoring the continued importance of developing new compounds with improved efficacy and safety profiles. In this context, five novel derivatives bearing the 4-(5-(naphthalene-2-yl)-2H-tetrazol-2-yl)-N-phenylbutanamide scaffold were synthesised through four known reactions: [3+2] cycloaddition, N-alkylation, saponification, and amidation. The purity of the compounds was confirmed by melting point analysis and thin-layer chromatography (TLC). Structural characterisation was achieved using infrared spectroscopy (IR), high-resolution mass spectrometry (HRMS), and proton and carbon nuclear magnetic resonance spectroscopy (1H-NMR and 13C-NMR). The acetylcholinesterase inhibitory activity of the synthesised compounds was evaluated at five concentrations to determine their IC50 values. Compound IVa exhibited the highest inhibitory potency, with an IC50 of 467.82±10.77 μM. In addition, the inhibitory potential of the compounds was analysed in relation to their physicochemical properties, including drug-likeness and blood-brain barrier permeability, with none violating Lipinski’s rule. Molecular docking studies further revealed that the naphthalene-containing framework demonstrated strong predicted binding affinity to AChE, which may contribute to the observed inhibitory activity. These findings provide novel and valuable structural insights for the development of AChE inhibitor derivatives as potential therapeutic strategies for Alzheimer’s disease.
In this study, ZnO nanomaterials were synthesised via a hydrothermal method and characterised using X-ray diffraction (XRD), scanning electron microscopy (SEM)/high-resolution transmission electron microscopy (HRTEM), Brunauer-Emmett-Teller (BET), analysis and diffuse reflectance spectroscopy (DRS). The obtained ZnO exhibited a wurtzite structure, high crystallinity, mesoporosity, a large specific surface area of 107.17 m2/g, and a band gap of 3.19 eV. Photocatalytic performance was comparatively evaluated under visible and UV irradiation using methylene blue (MB) and Rhodamine B (RhB) as model dyes. ZnO achieved >90-98% degradation, with consistently faster kinetics under UV irradiation and MB degrading faster than RhB. Pseudo-first-order fits yielded k(MB)=0.027 min-¹ (Vis) vs 0.076 min-¹ (UV) and k(RhB)=0.024 min-¹ (Vis) vs 0.066 min-¹ (UV), linking structureproperty features to light-dependent activity. These results establish a clear UV/Vis comparison and underscore the applicability of hydrothermal ZnO for wastewater treatment under practical illumination.
This study presents a robust control framework for enhancing real-time trajectory tracking of drones, overcoming the limitations of conventional proportional - integral - derivative (PID) controllers. The proposed architecture integrates a model predictive controller (MPC) with a Kalman filter to achieve accurate state estimation under noisy measurements, while explicitly handling system constraints. To address the computational bottleneck of MPC, we employ a fast gradient method-based solver with warm-start initialisation and fixed-iteration updates, ensuring deterministic and efficient execution. Experimental results on a Zynq-7000 FPGA-SoC demonstrate that the optimised software-only implementation achieves real-time feasibility, with an average computation time of 0.18 ms (≈1.4% of a 12.8 ms control cycle). Compared to PID, the proposed MPC reduces the maximum absolute axis tracking error from ≈0.5 to ≈0.2 m and generates significantly smoother control signals (~13x lower variance and =7x lower mean rate of change). These improvements enhance trajectory accuracy and flight stability while extending actuator lifetime, all achieved without requiring hardware acceleration.
Impurities whether organic, inorganic, or residual are unavoidable components that may arise during the synthesis, formulation, or storage of active pharmaceutical ingredients (APIs) and finished products. Even trace levels of these unwanted substances can significantly influence a drug’s efficacy, stability, and patient safety. Therefore, impurity profiling has become an integral part of modern pharmaceutical quality assessment, focusing on the identification, quantification, and control of impurities to ensure product reliability and regulatory compliance. Regulatory authorities across the globe, including the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use (ICH), the United States Food and Drug Administration (USFDA), and the European Medicines Agency (EMA), have established specific guidelines that define impurity thresholds, analytical validation requirements, and reporting standards. These frameworks help pharmaceutical manufacturers maintain consistent product quality and mitigate toxicological risks. The present review provides an overview of the regulatory aspects governing impurity profiling, outlines current global guidelines, and discusses the analytical and toxicological considerations essential for developing safe and compliant pharmaceutical products.
Thermoplastic cassava starch was plasticised with deep eutectic solvents (DES) based on choline chloride (ChCl), glycerol (Gly), and monoethanolamine (MEA) to balance free-volume creation and hydrogen-bond/ion-pair junctions. Binary (ChCl:Gly 1:3; ChCl:MEA 1:3) and ternary ChCl:Gly:MEA (1:1:1; 1:1:2; 1:2:1) DES at 25 wt% were melt-compounded and assessed by processing metrics, FTIR-ATR, Dynamic Mechanical Analysis (DMA), Scanning electron microscopy (SEM), X-ray diffractometry (XRD), wettability (Owens-Wendt-Rabel-Kaelble (OWRK)), and 56- day ambient aging (23±2°C, 50±5% RH). Gly-rich systems enhanced flow but depressed cohesion, whereas MEA-rich systems strengthened interchain junctions. The ternary ChCl:Gly:MEA 1:1:2 delivered the best process-structure-property balance, combining high melt processability with TS≈2.9 MPa and εb≈195% (n=5) and displaying the lowest 56-day ductility loss (Δεb≈-17%) compared with glycerol controls (≈-50%). FTIR (O-H/N-H red-shift and broadening), together with SEM fracture morphology and OWRK-derived surface energetics, supports a dual-mode plasticisation mechanism in which MEA-enabled junctions counterbalance Glyinduced free volume, thereby stabilising the amorphous phase and retarding retrogradation without sacrificing strength. This composition-structure map identifies ChCl:Gly:MEA (1:1:2) as a composition window that stabilises TPS mechanical performance during ambient aging.
Deep eutectic solvents (DESs) offer a green route to tailor starch–based plastics, yet how composition controls the interphase and aging in corn starch-lignin (CS–Lig) matrices remains unclear. Binary and ternary DESs were prepared from choline chloride (ChCl), glycerol (Gly), and monoethanolamine (MEA). In this study, CS–Lig was melt-processed with Gly (control) or with binary (ChCl:Gly 1:3; ChCl:MEA 1:3) and ternary ChCl:Gly:MEA DESs (1:1:1; 1:1:2; 1:2:1). We combined dynamic mechanical analysis (DMA), tensile testing, Fourier transform infrared (FTIR), Xray diffraction (XRD), scanning electron microscopy (SEM), and Owens–Wendt– Rabel–Kaelble (OWRK) wettability; we also integrated 0–56-day retrogradation assays. Relative to glycerol, DESs systematically enhanced performance: tensile strength (TS) rose to ≈ 2.97 MPa (CS–Lig–C.G.M 1:2:1) and elongation (εb) to ≈ 184 % (CS–Lig–C.G.M 1:1:2). One-way ANOVA (p < 0.001) and Tukey HSD separated ternary DESs from the control. FTIR showed broadened, red-shifted –OH/–NH envelopes; XRD indicated reinforced V-type order; SEM revealed tighter lignin encapsulation. Surface polarity was tuneable (γ ≈ 41–58 mN m⁻¹), and specific DESs mitigated aging-related losses. This composition–property map provides practical design rules to target strength/stiffness (e.g., 1:2:1) or ductility (e.g., 1:1:2) in greener CS–Lig bioplastics.
This study develops a Random Forest-based predictive framework to forecast course outcomes for Information Technology students at Binh Duong University using institutional academic records. Key features such as course codes, instructor identifiers, and encoded student IDs were combined with demographic and performance data through a standardised preprocessing pipeline that included feature engineering, normalisation, and class-imbalance handling. The final model achieved an accuracy of 0.98, a ROC AUC of 0.982, and a recall of 0.84 for the “Fail” class, demonstrating strong predictive power and practical value for early risk detection. Feature-importance analysis confirmed the dominant influence of course- and instructor-related variables alongside cumulative student histories, ensuring interpretability. By providing reliable and transparent predictions, the framework supports academic advisers in identifying at-risk students and designing timely interventions, highlighting the potential of institutional data for enhancing educational decision-making.
This article reviews and synthesises international experience and related studies on transport planning under the transit-oriented development (TOD) model at high-speed rail (HSR) and metro stations and compares them with existing policies and development plans in Ho Chi Minh city. Based on this foundation and combined with quantitative assessments and microsimulations of transport organisation around Phuoc Long station, the study proposes a set of TOD-oriented strategies for HSR and metro stations in the city. These include high-density and mixed-use urban planning, optimisation of public transport connectivity, promotion of green mobility, investment in modern transport infrastructure, development of high-quality public spaces, and improvements in traffic management. Simulation results comparing two scenarios (existing conditions and TOD-oriented transport planning) demonstrate that the TOD scenario yields significantly better performance. Projected passenger volumes increase substantially, vehicle speeds improve, and both delay and total travel time are reduced. These findings underscore the crucial role of TOD-oriented transport planning in enhancing the operational efficiency of public transport systems and improving the quality of the urban environment surrounding railway stations.
Longan (Dimocarpus longan Lour.) is a tropical fruit characterised by a short shelf life, which necessitates the application of postharvest preservation methods to reduce losses during distribution and consumption. A preservation mixture based on chitosan (CS) and the oligochitosan-iodine (OC-I2) complex was developed and evaluated for postharvest application in longan fruit storage. Ultraviolet-visible spectroscopy, X-ray diffraction, and Fourier-transform infrared spectroscopy confirmed that 2% H2O2 hydrolysis broke the β-1,4 glycosidic bond of CS without altering the OC structure. OC formed a complex with I2 through -NH2 and -OH groups. The chitosan-(oligochitosan--iodine) (CS-(OC-I2)) mixture at 9.00-22.5 g/l extended the shelf life of Xuong Com Vang longan fruits from 5 days (control) to 15-20 days at room temperature. Treated fruits showed reduced respiration rates, microbial density, weight loss, total soluble solids, browning index, and disease index, while maintaining greater fruit hardness and stabilising total soluble solids. These results indicate that the CS-(OC-I2) mixture is a promising preservative for longan and potentially other fruits, with the capacity to reduce postharvest losses during storage and distribution.
This article introduces an innovative system that integrates landmark identification with deep learning to enhance fall detection accuracy and reliability. By utilising advanced computer vision techniques, such as MediaPipe for spatial recognition, the system effectively differentiates between routine movements and actual falls. The integration of landmarks with a deep learning prediction algorithm minimises false alarms, ensuring timely responses to genuine falls. Comprehensive experimentation underscores the system’s versatility across various scenarios, emphasising its potential to improve safety and independence for older adults. The training process demonstrates a steady increase in accuracy, stabilising by the 40th cycle, while error rates decline significantly during the initial cycles. Real-time experiments, involving both male and female participants aged 8 to 50, recorded a remarkable 95% detection rate of falls, showcasing the system’s effectiveness and promising future applications in elder care and smart health monitoring environments.
The release of significant amounts of chemical detergents into the environment due to human activities raises concerns regarding environmental issues and human health. This study investigates the development of an eco-friendly dishwashing liquid derived from a fermented grapefruit peel solution, aiming to reduce organic waste and promote sustainable cleaning alternatives. The formulation involves fermenting a mixture of three parts grapefruit peel, one part brown sugar, and ten parts water in a sealed container for one month. The resulting solution effectively removes 89.2% of lard from plastic surfaces, eliminates 96.6% of bacteria, and produces foam comparable to that of commercial dishwashing liquids. Although no hydrolytic enzyme activity was detected using the well diffusion agar method with the respective substrates, the detergent effect of the fermentation broth likely arises from organic acids and biosurfactants, rather than from hydrolytic enzymes. The pH value of fermentation fluid decreased to 3 to 4.5 from the initial neutral pH. Consumer feedback highlights its potential as a safe, skin-friendly, and environmentally sustainable substitute for synthetic cleaners, contributing to broader environmental conservation efforts.
This article presents the development and application of SpectrumMate LR, a broadband, low-resolution spectrograph specifically tailored for use with small telescopes. SpectrumMate LR is designed to offer affordable and accessible spectroscopic capabilities for amateur astronomers, students, and educators, responding to the growing demand for versatile instrumentation in non-professional and educational settings. Employing a 300-grooves-per-millimetre diffraction grating together with 80 mm focal-length collimator and objective lenses, the system is optimised to analyse light across the entire visible spectrum. These optical parameters allow users to classify stars by spectral type, estimate stellar effective temperatures, and verify the transmission characteristics of astronomical filters. Laboratory and on-sky tests demonstrate SpectrumMate LR’s ability to record accurate, well-calibrated spectral data, validating its efficacy when observing a variety of both celestial and terrestrial light sources. Because all components are commercially available and assembly procedures are straightforward, the instrument fills an important niche for cost-effective spectroscopy, empowering a broader community to engage in detailed observational astronomy. By lowering financial and technical barriers, SpectrumMate LR promotes hands-on learning, encourages citizen-science contributions, and provides a practical pathway for schools, small observatories, and enthusiast clubs to expand their research and outreach activities. As a result, SpectrumMate LR stands out as an exemplary tool for introductory spectroscopy courses worldwide today.
