Search Results (252)

Effective scientific literature retrieval requires moving beyond surface-level term matching toward structured semantic reasoning. This paper presents a controlled empirical study of multi-stage retrieval for scientific literature, integrating lexical matching, dense semantic modeling, hybrid fusion, and cross-encoder re-ranking within a unified evaluation framework. The study is designed to analyze the interactions, trade-offs, and failure modes of these components in claim-based scientific search. Experiments on the SciFact benchmark demonstrate that dense models capture semantic similarity but remain insufficient when used in isolation. Hybrid fusion broadens the candidate pool but does not consistently outperform the best standalone dense retriever, as RRF-based fusion can dilute strong dense rankings when lexical and semantic signals diverge. Cross-encoder re-ranking proves to be the primary driver of final performance gains, with the best configuration, Hybrid (SciNCL + BM25) + Cross-Encoder, reaching NDCG@10 of 0.523, MAP@10 of 0.479, Recall@10 of 0.642, and MRR@10 of 0.497. Ablation analysis shows that lexical pseudo-relevance feedback (RM3) introduces query drift in claim-focused retrieval, and that passage-level max pooling weakens effectiveness by fragmenting document-level evidence. Cross-domain evaluation on SciFact, PubMedQA, and SciDocs demonstrates that the relative ranking of retrieval paradigms remains stable across datasets with varying difficulty levels, while also revealing that the RRF dilution effect intensifies on harder retrieval tasks. These findings suggest that effective scientific retrieval benefits from integrated multi-stage pipelines, and that understanding component-level interactions is essential for designing robust retrieval systems. Full article

Background/Objectives: An Endocrine Society Clinical Practice Guideline on vitamin D was published in 2024. Its main objective was the use of vitamin D to lower the risk of disease in individuals without established indications for vitamin D treatment or 25-hydroxavitamin D (25(OH)D) testing. The methodology followed the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) and the Evidence-to-Decision (EtD) framework. Evidence from randomized controlled trials (RCTs) retrieved by a systematic review was prioritized to inform this guideline. It was concluded that vitamin D supplementation reduces rickets and respiratory tract infections in children, mortality in individuals aged 75 years or older, pregnancy complications (outcomes), and progression of prediabetes to diabetes mellitus. Consequently, empiric vitamin D supplementation was recommended for individuals aged 1 to 18 years and ≥75 years, pregnant women, and individuals with prediabetes. Empiric vitamin D supplementation is defined as a vitamin D intake that exceeds the Dietary Reference Intakes (DRIs) and is implemented without 25(OH)D testing. Methods: This article provides a comprehensive guideline summary and critical appraisal based on a narrative review on scientific publications regarding that guideline. Results: Several publications discussed the 2024 Endocrine Society Clinical Practice Guideline on vitamin D. The main criticisms and discussion relate to unclear vitamin D dosages, guideline applicability to certain populations including controversy with previous vitamin D guidelines, and the implications of 25(OH)D testing. Conclusions: The 2024 Endocrine Society Clinical Practice Guideline on vitamin D followed a rigorous methodological approach with high quality standards but it leaves many open questions and uncertainties warranting clarification. Full article

Narrative extraction builds coherent ordered sequences of documents that trace how concepts develop over time, and is a growing area of information retrieval. In this work we focus on scientific literature, using a corpus of 3549 IEEE visualization research papers (1990–2022). A natural hypothesis is that augmenting embedding-based pathfinding with explicit domain knowledge should improve narrative quality. We present the Knowledge-Coherence Framework (KCF), which integrates structured metadata from OpenAlex into narrative extraction (building on the Narrative Trails algorithm), and conduct a systematic empirical investigation along three axes: (1) the effect of embedding model choice (MiniLM vs. SPECTER), (2) the effect of knowledge augmentation (with and without, plus sensitivity to the knowledge weight $\alpha$), and (3) the reliability of LLM-based evaluation (cross-agreement among 13 large language models). Throughout, mathematical coherence denotes the geometric mean of angular and topic similarity between consecutive documents along a path—an automatic, model-computed quantity inherited from Narrative Maps and Narrative Trails—while narrative quality refers to the LLM-judged construct. Using up to 600 evaluation pairs, we find that embedding model choice has a large effect on mathematical coherence (SPECTER: 0.94 vs. MiniLM: 0.81) and that, contrary to expectations, knowledge augmentation does not improve LLM-judged narrative quality—it slightly decreases it for both embeddings. Notably, the two notions dissociate: SPECTER produces the most mathematically coherent paths, yet MiniLM paths receive the highest LLM narrative-quality scores (5.87 vs. 5.36 out of 10). Alpha sensitivity analysis over five values ($\alpha \in \{0.0,0.3,0.5,0.7,1.0\}$, 500 pairs) confirms that LLM scores remain essentially flat while mathematical coherence steadily declines with increasing knowledge weight. Cross-model evaluation with 13 LLM judges shows high inter-model agreement (median Pearson $r=0.71$), supporting evaluation reliability. The main practical takeaways are that (i) embedding model choice, not knowledge augmentation, is the more consequential design decision, and (ii) mathematical coherence and LLM-judged narrative quality are distinct optimization targets that practitioners should not conflate. Full article

The accelerating growth of scientific publications has intensified the need for scalable and interoperable tools capable of supporting bibliometric analysis and research evaluation. In response to this challenge, this paper introduces IllustryFlow, a modular framework that combines n8n, an open-source workflow automation engine, with Illustry, a dynamic visualization platform, to extract, classify, and interpret scholarly data retrieved from OpenAlex. At the core of the framework is a multilingual BERT-based classification model implemented within the OpenAlex infrastructure, trained on the CWTS (Centre for Science and Technology Studies from Leiden University) classification schema and enriched with metadata features such as journal-level embeddings and citation graph information. IllustryFlow enables automated topic classification, clustering, and semantic visualization of citation networks, co-authorship structures, and thematic distributions. In this framework, Illustry and the custom n8n nodes represent components developed by the author, while OpenAlex and the OpenAlex-enhanced BERT model are integrated as external resources. The principal contribution of this study therefore consists of the architectural design and operational integration of these components into a unified, modular, automated, and reproducible bibliometric workflow. The proposed framework integrates an explicit and reproducible strategy for querying, semantic filtering, and selection of the bibliographic corpus. The framework was evaluated on a dataset of 1756 bibliographic records, and the entire workflow, including dashboard generation, was completed in approximately 90 s under the experimental conditions considered. The obtained results support the feasibility of the framework for scalable bibliometric workflows and indicate its practical potential for the analysis of heterogeneous bibliographic corpora while maintaining reproducibility under the analyzed conditions. Full article

Learning meaningful representations of scientific documents is essential for information retrieval, knowledge discovery, and recommendation systems. Traditional methods such as TF-IDF rely on lexical matching and fail to capture deeper semantic relationships, while transformer-based approaches typically depend on limited supervision signals. In this work, we propose a Triple-Source automatic supervision framework for learning document embeddings from scientific corpora. The model integrates three types of supervision–title–abstract pairs, same-category document pairs, and document-level semantic relationships—within a unified contrastive learning framework based on a multilingual XLM-RoBERTa encoder. Unlike prior approaches that rely on citation graphs or manual annotations, our method enables citation-free and annotation-free representation learning using only lightweight metadata. Experiments on a publicly available arXiv dataset consisting of 98,649 documents demonstrate improved semantic retrieval performance, achieving Recall@1 = 0.6181 for same-category retrieval and outperforming both TF-IDF and single-source transformer baselines. The learned embeddings also exhibit improved clustering of scientific domains, indicating more structured semantic representations. Full article

Driven by the global rise in animal protein demand, poultry farming has evolved into a highly intensive and technically complex sector. According to the FAO, animal protein production increased by about 16% in the past decade, with poultry alone expanding by 27% and becoming the leading source of animal protein. This intensification requires rapid, complex decisions across multiple aspects of production under uncertainty and strict time constraints. This study presents the development and evaluation of a conversational decision support system (DSS) designed to support decision-making to assist poultry producers, particularly broiler producers, in addressing technical queries across five key domains: environmental control, nutrition, health, husbandry, and animal welfare. As a proof-of-concept study, the reference context is intensive broiler production, covering common floor-rearing housing settings, including environmentally controlled and mechanically ventilated houses. The system combines a large language model (LLM) with retrieval-based generation (RAG) to ground responses in a curated corpus of scientific and technical literature. Additionally, it adds a reasoning component using Paraconsistent Annotated Evidential Logic Eτ, a non-classical logic designed to handle contradictory or incomplete information. Methodologically, Logic Eτ is used as a workflow-level control mechanism to gate clarification, domain routing, and answer adequacy signaling, rather than serving only as a post hoc label on generated outputs. Evaluation was conducted by comparing system responses with expert reference answers using semantic similarity (cosine similarity with SBERT embeddings). The results indicate that the system successfully retrieves and composes relevant content, while the paraconsistent inference layer makes results easier to interpret and more reliable in the presence of conflicting or insufficient evidence. These findings suggest that the proposed architecture provides a viable foundation for explainable and reliable decision support in modern poultry production, achieving consistent reasoning under contradictory or incomplete information where conventional RAG chatbots may produce unstable guidance. Full article

The Greater Caribbean manatee faces significant conservation challenges due to a lack of demographic data in low-visibility habitats. To address this, we present a refined automated manatee counting method pipeline integrating deep learning-based call detection with unsupervised individual counting. We resolved significant computational bottlenecks by implementing an offline feature extraction strategy, bypassing a 13-h processing lag for 43,031 audio samples. To mitigate overfitting in imbalanced bioacoustic datasets, non-parametric bootstrap resampling was employed to generate 100,000 balanced spectrograms. Benchmarking revealed that transfer learning via a VGG-16 backbone achieved a mean 10-fold cross-validation accuracy of 98.92% (±0.08%) and an F1-score of 98.08% for genuine vocalizations. Following detection, individual counting utilized k-means clustering on prioritized music information retrieval descriptors—spectral bandwidth, centroid, and roll-off—to resolve distinct acoustic signatures. This framework identified three individuals with a silhouette coefficient of 79.20%, demonstrating superior cohesion over previous benchmarks. These results confirm the automatic manatee count method as a robust, scalable framework for generating the scientific evidence required for regional conservation policies. Full article

The advancement of digital technologies such as photogrammetry, 3D scanning, Geographic Information Systems (GISs), and artificial intelligence has profoundly transformed archaeology by enabling more accurate documentation, analysis, and visualization of cultural heritage. These tools facilitate evidence preservation, enhance research processes, and broaden the possibilities for interpreting and disseminating archaeological knowledge. This scoping review synthesizes recent progress in the application of digital technologies for the reconstruction of archaeological evidence, emphasizing their main impacts on archaeological research while addressing existing challenges, limitations, and future perspectives, particularly focusing on the integration of artificial intelligence. A systematic review of the scientific literature was conducted using the PRISMA^® methodology, analyzing documents retrieved from databases such as Scopus, PubMed, IEEE Xplore, and ScienceDirect. One hundred and sixteen papers were selected, with a Cohen’s Kappa coefficient of 0.463 ensuring the reliability of the selection process. The findings reveal that the integration of digital technologies is redefining archaeological reconstruction methods and expanding the horizons of historical and heritage knowledge, requiring collaborative, ethical, and interdisciplinary approaches to achieve a more accurate, accessible, and sustainable archaeology in the future. Full article

The use of antibiotics as growth promoters in livestock production has contributed to the emergence and spread of antimicrobial resistance (AMR), posing a significant global public health threat specifically from the projected mortality burden. Although many countries have restricted the non-therapeutic use of antibiotics, practical and effective alternatives are still required to maintain livestock productivity. This scoping review examines the current evidence on non-antibiotic compounds evaluated as growth-promoting agents in livestock production. The primary objective of this search was to generate a comprehensive list of commonly applied alternatives to antibiotics used as growth promoters in livestock systems. A search was conducted in the CAB Abstracts, Web of Science Core Collection, and AGRICOLA databases. Prior to the scoping review, an initial list of alternatives to antibiotic components was generated through a screening of selected scientific sources and subsequently verified using Google Scholar for the period 2010–2025. This list included brief descriptions of each component, which were used to inform the keyword strategy for the scoping review. Eligible studies were screened in accordance with PRISMA-ScR guidelines, and data were extracted on compound type, livestock species, geographic region, and reported performance outcomes. The alternatives identified included probiotics and prebiotics, phytogenic compounds and essential oils, enzymes and organic acids, vaccines and immunostimulants, bacteriophages, and competitive exclusion products. A total of 1230 records were retrieved and imported into Zotero for reference management. After removal of duplicate records using Zotero’s built-in deduplication tool, 377 unique records remained for screening. Overall, these compounds demonstrated variable effects on feed efficiency, weight gain, and gut health. However, most studies were limited in scale, duration, and methodological consistency. As a result, comprehensive comparative trials and large-scale field evaluations are needed to support evidence-based policy recommendations and the sustainable implementation of alternatives to antibiotics in livestock production systems. Our findings identified six major categories that represent the most frequently reported alternatives to antibiotic growth promoters. Although probiotics, phytogenic, and organic acids were the most extensively studied, substantial heterogeneity in trial design, dosage, and production systems limited meaningful cross-comparisons. In addition, most studies focused on poultry and swine, with comparatively fewer investigations involving ruminant species. This scoping review was not intended to evaluate the efficacy or practical applicability of these alternatives; such assessments require further standardized and extensive studies before recommendations for their widespread application can be made. Full article

Frequent forest fires cause serious damage to ecosystems and socioeconomic systems, increasing the importance of fire prevention and risk assessment. Forest fuel is a fundamental determinant of forest fire behavior and a key component of fire risk management. However, a systematic synthesis of its global research evolution and emerging scientific challenges remains relatively insufficient. On the basis of 1257 publications retrieved from the Web of Science Core Collection (2010–2025) with the themes of “wildfire fuel” and “forest fuel,” this study employed CiteSpace for bibliometric analysis to systematically investigate research trends, collaboration patterns, and thematic evolution. The results show that forest fuel research has exhibited sustained growth overall, with notable peaks in 2016 and 2020, and reaching a historical high in 2023. The United States dominated both in publication output and institutional collaboration networks, forming a core research cluster together with Australia and Canada. Keyword co-occurrence and burst analyses revealed a shift in research hotspots—from early focus on forest fuel models and risk assessment at the wild–urban interface (WUI)—toward concerns about climate-change-driven fire seasonality, fuel moisture dynamics, and emergency response issues, reflecting the growing influence of climate change on wildfire patterns. Notably, this study identified several critical research gaps, including limitations in cross-regional integration of fuel moisture studies, insufficient attention to ignition prevention in WUI residential settings, and a lack of reproducible, open bibliometric workflows. By systematically mapping the knowledge structure and evolutionary trajectory of forest fuel research, this study provides a globally informed knowledge framework for the future advancement of forest fuel science and its deeper integration with forest fire management and policy making. Full article

Satellite-observed solar-induced chlorophyll fluorescence (SIF) provides a direct radiative link between solar radiation, photosystem de-excitation and vegetation photosynthetic activity. As multiple satellite missions now deliver global SIF products, machine learning (ML) has become a key tool for: (i) flexible nonlinear SIF retrieval, (ii) spatial reconstruction and downscaling of SIF fields, (iii) full-spectrum SIF reconstruction beyond narrow absorption windows, and (iv) data-driven analysis of the SIF–gross primary production (GPP) relationship. In addition, ML methods are increasingly used for: (v) uncertainty quantification (UQ) along the SIF information chain, and (vi) emulation (i.e., surrogate modelling) of radiative transfer models (RTMs) to accelerate computationally demanding SIF workflows. This review provides a conceptual and methodological survey of recent ML applications across the satellite SIF processing chain, summarises emerging products and methods, and highlights open challenges in uncertainty treatment, spectral reconstruction, and hybrid RTM–ML approaches. Particular emphasis is placed on the upcoming ESA FLEX mission, planned for launch in 2026, which will deliver multi-band SIF observations optimised for photosynthesis monitoring. While FLEX Level-2 (L2) operational processing will be based on physically grounded retrieval algorithms developed within ESA projects, ML is expected to play an important role in scientific exploitation and in the development of higher-level products (L3/L4), supporting high-resolution, uncertainty-aware SIF and GPP products and helping to bridge scales from leaf to ecosystem. Full article

With artificial intelligence (AI) rapidly increasing in popularity and presence in everyday life, new applications utilizing AI are being explored across virtually all domains, from banking and healthcare to cybersecurity to generative AI for images, voice, and video content creation. With that trend comes an inherent need for increased AI capabilities. One cornerstone of AI applications is the ability of generative AI to consume documents and utilize their content to answer questions, generate new content, correlate it with other data sources, and more. No longer constrained to text alone, we now leverage multimodal AI models to help us understand visual elements within documents, such as images, tables, figures, and charts. Within this realm, capabilities have expanded exponentially from traditional Optical Character Recognition (OCR) approaches towards increasingly utilizing complex AI models for visual content analysis and understanding. Modern approaches, especially those leveraging AI, are now focusing on interpreting more complex diagrams such as flowcharts, block diagrams, Unified Modeling Language (UML) diagrams, electrical schematics, and timing diagrams. These diagram types combine text, symbols, and structured layout, making them challenging to parse and comprehend using conventional techniques. This paper presents a historical analysis and comprehensive survey of scientific literature exploring this domain of visual understanding of complex technical illustrations and diagrams. We explore the use of deep learning models, including convolutional neural networks (CNNs), recurrent neural networks (RNNs), and transformer-based architectures. These models, along with OCR, enable the extraction of both textual and structural information from visually complex sources. Despite these advancements, numerous challenges remain, however. These range from hallucinations, where the content extraction system produces outputs not grounded in the source image, which leads to misinterpretations, to a lack of contextual understanding of diagrammatic elements, such as arrows, grouping, and spatial hierarchy. This survey focuses on five key diagram types: flowcharts, block diagrams, UML diagrams, electrical schematics, and timing diagrams. It evaluates the effectiveness, limitations, and practical solutions—both traditional and AI-driven—that aim to enable the extraction of accurate and meaningful information from complex diagrams in a way that is trustworthy and suitable for real-world, high-accuracy AI applications. This survey reveals that virtually all approaches struggle with accurately extracting technical diagram information. It also illustrates a path forward. Pursuing research to further improve their accuracy is crucial for supporting and enabling various applications, including complex document question answering and Retrieval Augmented Generation (RAG), document-driven AI agents, accessibility applications, and automation. Full article

Itaúba (Mezilaurus itauba (Meisn.) Taub. ex Mez) is an Amazonian forest tree whose high-quality timber has driven sustained commercial exploitation, leading to its classification as threatened with extinction. This systematic scoping review synthesizes the current scientific knowledge on M. itauba. A systematic search of the Web of Science, Scopus, and SciELO databases retrieved studies published in English, Portuguese, and Spanish. Sixty-eight articles were analyzed using quantitative and qualitative approaches. Publications were concentrated between 2012 and 2025, largely derived from research conducted in Brazil and disseminated mainly through national journals. Overall, the literature is dominated by studies on wood technological properties, whereas research on the ecology and silviculture of M. itauba remains limited and often methodologically insufficient to support effective conservation actions. Based on the synthesis of identified knowledge gaps, we highlight as research priorities (i) the generation of empirical data on field performance across developmental stages, from nursery based seedling production to establishment and growth under open field and managed forest conditions; (ii) advancement of knowledge on genetic attributes, including structure and adaptive potential, to support conservation strategies and the selection of planting material; and (iii) integration of ecological interactions, ecophysiological responses, and regeneration processes into applied management frameworks capable of informing evidence based public policies. Addressing these priorities is essential to support conservation planning and the sustainable management of M. itauba. Full article

Background: Social media, especially Instagram, spreads nutrition-related information that often lacks scientific rigor. Many women report feeling inadequately informed about women’s health by healthcare professionals, turning to social media, increasing exposure to misinformation. Objectives: The NutriWomen platform aims to assess the quality, methodological soundness, and credibility of nutritional health claims and dietary recommendations on Instagram targeting women across different life stages. Its goal is to develop a systematic and scientifically grounded evaluation framework to assess Instagram nutrition-related claims and the methodological quality and interpretability of their supporting evidence, and to translate the results into accessible outputs that help women make informed nutrition decisions across life stages. Methods: This study follows a five-stage design Stage 1 involves a retrospective content analysis of Instagram posts containing nutrition-related claims targeted at women, identified through the “Top posts” function and screened using predefined criteria. Stage 2 assesses information quality using a validated 14-item tool. Stage 3 evaluates the scientific accuracy of claims by formulating PI(E)CO(TS) questions, selecting key outcomes, retrieving evidence from PubMed and the Cochrane Database, and appraising systematic reviews with a modified AMSTAR-2 tool incorporating GRADE ratings, when available. Stage 4 develops the NutriWomen website platform to translate assessments into accessible visual summaries. Stage 5 conducts a mixed-methods study with peri-, meno-, and postmenopausal women to explore information needs and evaluate platform usability through focus groups. Conclusions: The NutriWomen platform will be the first website to systematically publish the results of evaluations assessing the scientific quality of nutritional health claims on Instagram targeted at women across different life stages. It will provide a replicable methodology, and a digital tool designed to empower women with trustworthy nutrition information, with the potential to enhance health literacy and promote better health outcomes. Full article

Timely and accurate inspection of concrete bridges is critical to ensuring structural integrity and public safety. Traditional visual inspections conducted by human inspectors are labour-intensive, inconsistent, and often limited in their ability to access all structural components, particularly in hazardous or inaccessible areas. Image-based inspection techniques have emerged as a safer and more efficient alternative, and recent advancements in deep learning have significantly enhanced their diagnostic capabilities. This systematic review critically evaluates 77 studies that applied deep learning approaches to the detection and classification of surface defects in concrete bridges using 2D images. Relevant publications were retrieved from major scientific databases, screened for eligibility, and analyzed in terms of model type, training strategies, and evaluation metrics. The reviewed works encompass a wide spectrum of algorithms—spanning classification, object detection, and image segmentation models—highlighting their architectural features, strengths, and trade-offs in terms of accuracy, computational complexity, and real-time applicability. Key findings reveal that transfer learning, data augmentation, and careful dataset composition are pivotal in improving model performance. Moreover, the review identifies emerging research trajectories, such as integrating deep learning with Building Information Modeling (BIM), leveraging edge computing for real-time monitoring, and developing rich annotated datasets to enhance model generalizability. By mapping the current state of knowledge and outlining future research directions, this study provides a foundational reference for researchers and practitioners aiming to deploy deep learning technologies in bridge inspection and infrastructure monitoring. Full article