Search Results (1,778)

The acid dissociation constant (pK_a) is a fundamental parameter governing the environmental fate of organic compounds. Accurate pK_a prediction remains challenging, as traditional binary Morgan fingerprints (B-MF) fail to capture stoichiometric information critical for modeling substituent effects. This study developed an interpretable machine learning framework for pK_a prediction by integrating count-based Morgan fingerprints (C-MF) with ensemble algorithms. Through systematic comparison across four algorithms (Catboost, XGBoost, GBDT, RF), C-MF consistently outperformed B-MF due to its ability to quantify functional group multiplicity. Subsequent SHAP-based recursive feature elimination (SHAP-RFE) optimized the model, identifying Catboost with only 81 features as the optimal architecture, achieving a test-set R² of 0.890 and RMSE of 1.026. SHAP analysis revealed that the model’s decisions are driven by chemically intuitive features, forming a hierarchical framework where primary ionizable sites set the baseline pK_a and electronic modifiers fine-tune it. The applicability domain, defined using the AD_SAL method, yielded high-confidence predictions (R² = 0.926). External validation on an independent open-source dataset containing 6876 acidic compounds, combined with results from AD_SAL application domain characterization, enabled accurate pK_a prediction for 390 compounds within the application domain (R² = 0.890, RMSE = 0.942). This further confirms the model’s strong generalizability. This work provides a robust and generalizable tool for high-performance pK_a prediction, with significant potential for applications in environmental risk assessment. Full article

Voids behind the lining that develop during long-term operation can seriously compromise the seismic safety performance of metro shield tunnels. To investigate the influence of such void defects on large-diameter shield tunnels, this study systematically analyzed the causes and distribution patterns of voids. A three-dimensional discontinuous finite element model was developed to simulate the interaction among lining segments, connecting bolts, and surrounding rock. The seismic responses, including circumferential stress, interface slip, interface opening, and bolt tensile stress, were analyzed considering coupled factors such as the void circumferential angle, radial depth, distribution location, and geological conditions. Single-factor and multi-factor sensitivity analyses were conducted to evaluate the significance of the above coupled factors on the overall seismic response. The results show that lining circumferential stress, displacement, interface opening, and bolt stress increase with void enlargement, a shift in void location from the crown to the haunch, and deterioration of geological conditions. A void located at the right haunch leads to a peak circumferential stress of 3.27 MPa, causing local segment damage. Sensitivity analysis reveals that void location is the most influential factor affecting the seismic response, while geological conditions exhibit lower sensitivity. A predictive model for the peak circumferential stress around the void was established using multiple linear regression, incorporating void position, circumferential angle, and radial depth. Within the parameter range considered in this study, this model provides a theoretical basis and practical reference for rapid seismic risk assessment and safety management of shield tunnels with void defects. Full article

Due to climate change, seedling damage caused by drought stress is expected to increase in both afforestation sites and nurseries. Therefore, to ensure stable seedling production under high-temperature conditions and to cultivate seedlings with enhanced drought tolerance through hardening treatments, the development of an effective irrigation system is required. Conventional physiological methods for non-destructive drought detection, such as chlorophyll fluorescence and leaf temperature measurements, require expensive and manual operation, thereby limiting their real-time applicability in forest nurseries. This study evaluated the applicability of using image-based leaf angle measurements for drought stress detection in Quercus acutissima Carruth. seedlings. One-year-old seedlings were grown under two water regimes—well-watered (CT: control) and unwatered (DT: drought)—through Day 8. Statistical analyses (RMANOVA) revealed that changes in the leaf angle parameter PMD–MD (the difference between the previous and current measurement days) showed treatment effects similar to those of the physiological responses ΦNO (quantum yield of non-regulated energy dissipation) and qL (fraction of open PSII reaction centers) to drought on Day 6. Leaf angle reflected drought stress but did not precede physiological changes, indicating its role as a complementary rather than an early indicator. Multiple regression models identified AT (air temperature), SM (soil moisture), Fm′ (maximum fluorescence in the light-adapted state), and VPD (vapor pressure deficit) as the main factors influencing leaf angle variation. Although leaf angle was affected by combined environmental stresses such as high temperature, it was less sensitive to heat stress than physiological responses based on RMANOVA results. These results indicate the potential of image-based leaf angle measurements for drought stress detection. To establish plant-based smart irrigation systems, future studies should validate and refine this approach using larger datasets. Full article

While sustainability is set as a goal by a broad range of international organizations, its definition varies, and there is still a lack of practical criteria for product designers to evaluate the degree of (un)sustainability in the design phase. Life cycle assessment (LCA) can allow quantification of the environmental impacts of a product but is often carried out post-design, when the manufacturing process is already settled. Finally, while significant advances have been made towards standardizing LCA calculations by providing product category rules, large uncertainties remain in the calculation results due to a lack of transparency regarding the choices of databases, system boundaries, allocation, cut-off rules, and level of data granularity. A practical way to improve in those areas is to share with the semiconductor community a parametrizable life cycle inventory (LCI) model based on a target device to (1) identify knowledge gaps in LCA methods for such products, (2) identify the main process variables, and (3) provide a starting point for LCA calculations by the designers themselves. With this aim, a parametrizable cradle-to-gate manufacturing LCI model was developed based on the peer-reviewed process flow of a trench field-stop silicon insulated gate bipolar transistor (IGBT) semiconductor power device. The model allows computation of the environmental impacts of the IGBT manufacturing process based on different tunable parameters such as die size, wafer diameter, manufacturing yield, abatement efficiency, wafer fab throughput, wafer fab location, and associated electricity mix. Embedding a high level of data granularity, it helps identify, at elementary process levels, key environmental hotspots and associated technical levers for their reduction. Analysis of the IGBT manufacturing process tends to demonstrate the importance of an impact assessment approach considering multiple environmental categories, going beyond the sole focus on greenhouse gas emissions and accounting for potential transfers of impact. With an open-source mindset and in a continuous improvement prospective, the manufacturing inventory model and its associated tools are freely available from a public GitHub repository and open for comments and consolidation from users. Full article

Hybrid railway assets such as workshops and depots combine building, mechanical, electrical and plumbing (MEP)/industrial, and linear infrastructure domains, increasing coordination complexity and challenging continuity from the Project Information Model (PIM) to the Asset Information Model (AIM). Although Employer’s Information Requirements (EIR), Asset Information Requirements (AIR), and the BIM Execution Plan (BEP) prescribe deliverables and processes, a persistent gap remains between documentary prescriptions and the auditable evidence needed to support traceable decisions within the Common Data Environment (CDE). This paper proposes an ISO 19650-aligned governance framework that operationalizes the EIR/AIR → BEP → CDE transition by: (i) structuring the asset using Functional Units (FUs) as a stable anchor for PIM → AIM continuity; and (ii) implementing a pre-Published Quality Gate that separates control into three non-substitutable dimensions (spatial, semantic, and data). The approach is implemented as a tool-neutral, reproducible workflow (inputs → checks → outputs → publish) and produces a minimal, persistent evidence package in the CDE (file-level report, package summary, publish/hold decision record, and Nonconformity Report (NCR)/BIM Collaboration Format (BCF) traceability), with explicit roles governing the Shared → Published transition. Across 22 Industry Foundation Classes (IFC), deliverables from two depot cases and multiple delivery states, All Gates Pass ranged from 25.0% to 44.4% depending on Case × State; overall, 14/22 deliverables (63.6%) would be held pending correction under the gate. Although validated on Spanish railway depots, the framework is grounded in ISO/openBIM standards and is designed for transferability to other international contexts and complex asset types where multidisciplinary federation and PIM → AIM continuity pose similar challenges. Full article

The automatic detection of foliar nutritional deficiencies through computer vision represents a promising alternative within precision agriculture practices, reducing dependence on laboratory analyses and the subjectivity associated with visual inspection. This systematic review maps and compares the application of machine learning (ML) and deep learning (DL) techniques to nutritional diagnosis across different crops, highlighting methodological trends, barriers to model adoption under field conditions, and existing research gaps. Following the PRISMA guidelines (PRISMA-P and PRISMA-2020), searches were conducted in the Scopus, IEEE Xplore, and Web of Science databases, using a defined time frame and explicit inclusion and exclusion criteria, resulting in 200 articles included (2012–2026; last search on 2 February 2026). The results indicate a predominance of DL-based approaches and RGB imagery, with applications concentrated in crops such as rice and in macronutrients, mainly nitrogen (N), phosphorus (P), and potassium (K), and report a marked increase in publications from 2020 onward. Although many studies report high performance, the evidence is largely derived from controlled environments and proprietary datasets, which limit model comparability, reproducibility, and generalization to real-world scenarios. Accordingly, the main research gaps include limited validation under field conditions, identified as the primary practical barrier; the underrepresentation of micronutrients and multiple-deficiency diagnosis; and the need for lightweight architectures suitable for deployment in mobile and edge-computing applications. It is concluded that ML and DL techniques offer promising alternatives for automated nutritional diagnosis; however, advances in data standardization, open-access datasets, and validation under real field conditions are essential for consolidating these technologies in practical applications. Full article

Medical question answering (QA) systems and conversational agents have attracted growing interest as tools that can assist clinicians, support medical students, and help patients navigate complex information sources. However, existing evaluations of retrieval strategies largely overlook the cost–benefit balance—here referred to as frugality, under realistic computational constraints. This work introduces a frugality-based evaluation framework that jointly assesses accuracy improvements and computational cost to determine when retrieval-augmented generation is beneficial in medical question answering, rather than evaluating retrieval effectiveness through accuracy alone. This study addresses these gaps through a systematic comparative framework that evaluates retrieval relevance, computational efficiency, and knowledge base composition across multiple biomedical QA tasks. We employ open-source LLMs (LlaMA-3-8B-Instruct, Mistral-7B-Instruct-v0.3, and DeepSeek-7B-Chat) across three benchmark medical QA datasets, including MedMCQA, MedQA-USMLE, and PubMedQA. In addition to that, we evaluate a dataset with larger contexts to simulate model distraction across the CliniQG4QA dataset using additional models (Meditron-7B, Qwen2.5-7B-Medical, Medgemma-4B, Phi-3-mini-4k-Instruct, and GPT4o-Mini). We examine how retrieval design choices alter the accuracy–latency trade-off, examining how relevance, corpus design, and hardware constraints interact in medical retrieval-augmented generation (RAG) systems. Our comprehensive results demonstrate when retrieval is genuinely beneficial versus when it imposes unnecessary computational costs, highlighting interactions between relevance and corpus designs in QA. Full article

Smart city decision-making increasingly relies on heterogeneous urban data sources. Dense traffic sensor streams provide continuous quantitative measurements, while citizen-generated textual reports offer event-driven contextual information. However, integrating these modalities remains challenging due to temporal misalignment, textual sparsity, and semantic noise. This paper investigates multi-modal learning for traffic congestion severity prediction through an experimental integration of open traffic sensor data (METR-LA: Los Angeles, USA) and citizen-generated textual reports (NYC 311: New York City, USA). Congestion severity is formulated as a four-class classification task derived from traffic speed measurements. We propose an end-to-end framework that combines: (i) sensor time-series encoding using a GRU-based temporal encoder, (ii) textual representation learning using a BERT-based encoder, (iii) a symmetric time-window alignment strategy (±Δ) to associate irregular reports with sensor time steps, and (iv) multiple fusion architectures, including early fusion, late fusion, and a cross-attention module for cross-modal interaction modeling. Experiments on publicly available datasets show that multi-modal early fusion achieves the best overall performance (Accuracy = 0.8283, Macro-F1 = 0.8231) compared to uni-modal baselines. In the studied cross-city setting with sparse and weakly aligned textual signals, the proposed cross-attention fusion does not outperform the strong sensor-only baseline, suggesting that the sensor modality dominates when cross-modal signal strength is limited. These results highlight both the potential and the practical constraints of multi-modal fusion in heterogeneous smart-city environments, emphasizing the importance of alignment design, modality relevance, and transparent experimental validation. Full article

This study investigates the macroeconomic impact of financial corruption and institutional weakness on Pakistan’s economy from 1996 to 2023, addressing a critical research gap in quantifying the simultaneous effects of shadow economy operations and poor governance on economic growth. Grounded in institutional economics theory, the research tested hypotheses that weak control of corruption and a large shadow economy negatively affect GDP growth, while also examining the roles of tax revenue, inflation, trade openness, and foreign direct investment. Utilizing a dual-methodological approach, this study employed multiple regression analysis with stationary testing to ensure robust inference, complemented by Random Forest machine learning with Leave-One-Out Cross-Validation for predictive accuracy and variable importance ranking. The econometric results identified shadow economy size and inflation rate as the most statistically significant barriers to growth, with a one percentage point increase in each associated with 0.32 and 0.08 percentage point reductions in GDP growth, respectively (p < 0.05). Control of corruption and institutional quality showed positive but statistically weaker effects. The machine learning analysis corroborated these findings, ranking shadow economy (31.8%) and inflation (24.5%) as the dominant predictors of GDP growth, with the Random Forest model achieving superior predictive performance (R² = 0.68) compared to traditional linear regression (R² = 0.45). Both techniques converged on the conclusion that formalizing informal activity and stabilizing prices represent the most impactful policy levers for growth enhancement, while institutional quality improvements operate through indirect channels. The findings underscore the urgent need for policymakers to prioritize inflation control through credible monetary policy and to formalize informal economic activity via simplified regulations and anti-corruption measures. This research provides a replicate dual-methodology framework for analyzing institutional economic issues in developing nations with limited data. Full article

The development of passive microwave sensors traces back to Robert Dicke’s pioneering experiments in the 1940s. Since then, microwave radiometry has evolved into a key tool for Earth observation, strengthened by data from multiple satellite missions operating across different wavelengths. This paper reviews the state of the art in microwave radiometry for monitoring land surfaces. After introducing the theoretical foundations underpinning current missions, we present an overview of major satellite instruments. We then examine early theoretical advances in retrieving soil moisture and snow properties, two applications that contributed to the future development of satellite microwave radiometry missions for the observation of surface variables. Particular attention is given to radiative transfer theory and its solutions, which model the effects of roughness, vegetation, and snow cover. These approaches form the basis of today’s retrieval algorithms and remain central to future missions. Subsequent sections highlight the use of passive microwave data for estimating a variety of surface variables, the role of passive microwave in data assimilation systems and forthcoming missions dedicated to land monitoring. The review concludes with key achievements, ongoing challenges, and open issues—such as soil moisture retrieval under dense vegetation or snow property retrieval in melting conditions. Addressing these limitations is critical to fully exploiting microwave radiometry in the context of climate research and mitigation strategies. Full article

Augmented Reality (AR) and Large Language Models (LLMs) have made significant advances across many fields, opening new possibilities, particularly in complex machine operations. In complex operations, non-expert users often struggle to perform high-precision tasks and require constant supervision to execute tasks correctly. This paper proposes a novel AR-MLLM-based training system that integrates AR, multimodal large language models (MLLMs), and prompt engineering to interpret real-time machine feedback and user activity. It converts extensive technical text into structured, step-by-step commands. The system uses a prompt structure developed through an iterative design method and refined across multiple machine operation scenarios, enabling ChatGPT to generate task-specific contextual digital overlays directly on the physical machines. A case study with participants was conducted to assess the effectiveness and usability of the AR-MLLM system in Coordinate Measuring Machine (CMM) operation training. The experimental results demonstrate high accuracy in task recognition and feature measurement activity. The data further show reduced time and user workload during task execution with the proposed AR-MLLM system. The proposed system not only provides real-time guidance and enhances efficiency in CMM operation training but also demonstrates the potential of the AR-MLLM design framework for broader industrial applications. Full article

Retrieval-Augmented Generation (RAG) has been observed to encounter challenges in heterogeneous query scenarios characterised by varying evidence requirements and reasoning depths. In order to address this limitation, the present paper puts forward a proposal for an Adaptive Multi-Source RAG framework (AMSRAG) that integrates query complexity awareness with confidence-aware fusion. The framework performs query complexity classification with a pretrained language model, calibrates the classification confidence to guide the dynamic scheduling of retrieval paths and the adjustment of fusion weights, and enables a controllable balance between answer quality and retrieval efficiency through hierarchical path selection and cross-source weighting. The experiments conducted on multiple open-domain question-answering datasets demonstrate that the query complexity classifier achieves an accuracy of 85.9% and a Macro-F1 score of 85.4%. These outcomes indicate the potential for the classifier to generate a reliable decision signal, which can subsequently be utilised to guide the process of adaptive retrieval and fusion. The proposed framework demonstrates a marked improvement in terms of both answer accuracy and retrieval relevance when compared to the fixed-pipeline RAG. In scenarios involving high-confidence queries, the system has been shown to effectively avoid redundant retrieval, thereby reducing the average number of retrievals. In instances of low-confidence complex queries, the system has been shown to enhance evidence coverage and completeness of answers through multi-source retrieval and confidence-weighted fusion. This study proposes a novel methodology for enhancing the adaptability and resource efficiency of RAG systems in response to heterogeneous query conditions. Full article

Personalized e-learning seeks to adapt sequences of learning activities to individual learners, yet most existing adaptive platforms continue to rely on heuristic rules or single-objective optimization strategies. This paper introduces AlphaLearn, a conceptual evolutionary agent that frames learning pathway design as a constrained multi-objective optimization problem. The framework integrates knowledge graphs, learner modelling, and evolutionary algorithms to generate, evaluate, and iteratively refine candidate learning pathways under multiple pedagogical criteria. The contribution of this work is threefold. First, it presents a structured architectural framework for evolutionary learning pathway optimization, including a formal description of the optimization cycle and pathway representation. Second, it provides a descriptive analysis of large-scale learning analytics data from the Open University Learning Analytics Dataset (OULAD), illustrating substantial variability in learner outcomes, failure rates, and dropout across modules. Third, it offers an explicit discussion of fairness and bias mitigation, positioning equity as an integral dimension of adaptive pathway optimization rather than a post-hoc concern. The descriptive findings highlight pronounced heterogeneity in learner performance and engagement, motivating the need for adaptive systems capable of balancing learning effectiveness, efficiency, engagement, and fairness. While AlphaLearn is presented as a conceptual and methodological framework rather than a validated system, it establishes a foundation for future empirical evaluation and the development of fairness-aware evolutionary approaches to personalized e-learning. Full article

Globalization has reorganized industrial spatial patterns, embedding regional economies into complex global production systems. However, the existing literature primarily focuses on the national level, leaving the “global-national-local” multi-scale linkages of sub-national regions underexplored. Focusing on Guangdong, which is China’s most open economic gateway, this study constructs a nested Multi-Regional input–output (MRIO) model to systematically reveal its industrial linkage paths across multiple scales. The results demonstrate that Guangdong features “strong local services and extensive global connections.” Specifically, the network is led by the high-R&D-intensity category and supported by energy and low-R&D categories, highlighted by two core supply paths, which are non-metallic mineral supply for construction and metal product support for optical–electrical manufacturing. Four heterogeneous modes are identified: resource security, innovation-driven dual circulation, cost-competitive regional division, and export-oriented service support. Crucially, the provincial “domestic intermediate chains plus international core chains” logic underscores Guangdong’s role as a bridge connecting Global and Domestic Value Chains. Theoretically, this work enriches the local dimension of Global Production Network theory. Methodologically, it provides an operational tool for nested analysis. Practically, it offers policy evidence for open economies to optimize industrial layouts and enhance supply chain resilience. Full article

Background: Machine learning (ML) has been widely adopted in decision-making, making fairness a central ethical and scientific priority. We developed the Themis chatbot, a Large Language Model (LLM) system designed to explain concepts of ML fairness in an accessible, conversational format. Methods: The development followed four stages: (1) curating a document corpus of 286 peer-reviewed publications on ML fairness; (2) development of Themis by combining a modern LLM (OpenAI’s GPT-4o) with Retrieval Augmented Generation (RAG); (3) creation of a 340-item benchmark dataset, the FairnessQA; and (4) evaluating performance against state-of-the-art non-augmented LLMs (DeepSeek R1, GPT-4o, GPT-5, and Grok 3). Results: For the multiple-choice questions, Themis achieved an accuracy of 96.7%, outperforming DeepSeek R1 (90.0%), GPT-4o (89.3%), GPT-5 (92.0%), and Grok 3 (86.7%), and the overall difference was statistically significant (χ²(4) = 10.1, p = 0.038). In the closed-ended questions, Themis achieved the highest accuracy (96.7%), while competing models ranged from 78.0% to 84.0%, and the overall difference was significant (χ²(4) = 23.9, p < 0.001). In the open-ended questions, Themis achieved the highest mean scores for correctness (M = 4.62), completeness (M = 4.59), and usefulness (M = 4.56), and differences were statistically significant (correctness: F(4, 195) = 20.91, p < 0.001; completeness: F(4, 195) = 7.76, p < 0.001; usefulness: F(4, 195) = 2.90, p < 0.001). By consolidating scattered research into an interactive assistant, Themis makes fairness concepts more accessible to educators, researchers, and policymakers. This work demonstrates that retrieval-augmented systems can enhance the public understanding of machine learning fairness at scale. Full article