Frontiers in Big Data最新文献

Introduction: The adoption of Large Language Models (LLMs) in search systems necessitates new evaluation methodologies beyond traditional rule-based or manual approaches.

Methods: We propose a general framework for evaluating structured outputs using LLMs, focusing on search query parsing within an online classified platform. Our approach leverages LLMs' contextual reasoning capabilities through three evaluation methodologies: Pointwise, Pairwise, and Pass/Fail assessments. Additionally, we introduce a Contextual Evaluation Prompt Routing strategy to improve reliability and reduce hallucinations.

Results: Experiments conducted on both small- and large-scale datasets demonstrate that LLM-based evaluation achieves approximately 90% agreement with human judgments.

Discussion: These results validate LLM-driven evaluation as a scalable, interpretable, and effective alternative to traditional evaluation methods, providing robust query parsing for real-world search systems.

This review systematically summarizes recent advances in microarray feature selection techniques and their applications in biomedical research. It addresses the challenges posed by the high dimensionality and noise of microarray data, aiming to integrate the strengths and limitations of various methods while exploring their applicability across different scenarios. By identifying gaps in current research, highlighting underexplored areas, and proposing clear directions for future studies, this review seeks to inspire academics to develop novel techniques and applications. Furthermore, it provides a comprehensive evaluation of feature selection methods, offering both a theoretical foundation and practical guidance to help researchers select the most suitable approaches for their specific research questions. Emphasizing the importance of interdisciplinary collaboration, the study underscores the potential of feature selection in transformative applications such as personalized medicine, cancer diagnosis, and drug discovery. Through this review, not only does it provide in-depth theoretical support for the academic community, but also practical guidance for the practical field, which significantly contributes to the overall improvement of microarray data analysis technology.

Community health outcomes significantly impact older populations' wellbeing and quality of life. Traditional analytical methods often struggle to accurately predict health risks at the community level due to their inability to capture complex, non-linear relationships among various health determinants. This study employs a Random Forest Algorithm (RFA) to address this limitation and enhance the predictive modeling of community health outcomes. By leveraging ensemble learning techniques and multi-factor analysis, this study aims to identify and quantify the relative contributions of key health indicators to risk assessment. The study begins with comprehensive data collection from diverse health sources, followed by a systematic preprocessing stage, which includes resolving missing values, normalizing variables, and encoding categorical features. Using bootstrap sampling, multiple decision trees were trained on random subsets of health data, ensuring variability in the model learning. The trees grow to full depth and aggregate their predictions to enhance the accuracy. An out-of-bag (OOB) error estimation was applied to refine the model and provide unbiased performance assessments, ensuring robust generalization to unseen data. The proposed model effectively analyzes key health indicators, ranking the feature importance to determine the most influential predictors of health risks. Results indicate that RFA achieves an accuracy rate of 92%, outperforming conventional prediction methods in terms of precision and recall. These findings underscore the efficacy of Random Forest in identifying critical health risk factors, paving the way for targeted and data-driven public health management strategies and interventions tailored to older adults.

Introduction: The widespread emergence of deepfake videos presents substantial challenges to the security and authenticity of digital content, necessitating robust detection methods. Deepfake detection remains challenging due to the increasing sophistication of forgery techniques. While existing methods often focus on spatial features, they may overlook crucial temporal information distinguishing real from fake content and need to investigate several other Convolutional Neural Network architectures on video-based deep fake datasets.

Methods: This study introduces an RLNet deep learning framework that utilizes ResNet and Long Short Term Memory (LSTM) networks for high-precision deepfake video detection. The key objective is exploiting spatial and temporal features to discern manipulated content accurately. The proposed approach starts with preprocessing a diverse dataset with authentic and deepfake videos. The ResNet component captures intricate spatial anomalies at the frame level, identifying subtle manipulations. Concurrently, the LSTM network analyzes temporal inconsistencies across video sequences, detecting dynamic irregularities that signify deepfake content.

Results and discussion: Experimental results demonstrate the effectiveness of the combined ResNet and LSTM approach, showing an accuracy of 95.2% and superior detection capabilities compared to existing methods like EfficientNet and Recurrent Neural Networks (RNN). The framework's ability to handle various deepfake techniques and compression levels highlights its versatility and robustness. This research significantly contributes to digital media forensics by providing an advanced tool for detecting deepfake videos, enhancing digital content's security and integrity. The efficacy and resilience of the proposed system are evidenced by deepfake detection, while our visualization-based interpretability provides insights into our model.

In emergency situations, 9-1-1 is considered the first point of contact, and their call handlers play a crucial role in managing the emergency response. Due to the large number of daily calls and the hectic routine, there are severe chances that the call handlers can make any mistake or human error during data taking in a high-pressure environment. These mistakes or errors impact 9-1-1 performance in emergencies. To address this problem, this research introduces an AI-powered digital health framework called Emergency Calls Assistant (ECA) that leverages artificial intelligence (AI) and natural language processing (NLP) techniques to assist call handlers during data collection. ECA is designed to predict the type of emergency, suggest relevant questions to collect deeper information, suggest pre-arrival instructions to emergency personnel, and generate incident reports that helps in data-driven decision making. The ECA framework works in two phases; the first phase is to convert the audio call into digital textual form, and the second phase is to analyze the textual information using NLP tools and mining techniques to retrieve contextual information. The second phase also deals with emergency categorization using a support vector machine (SVM) learning model to prioritize the emergency dealing with an accuracy of 92.7%. The key factors involved in categorization by ML models are the severity of injury and weapons involvement. The objective of ECA's development is to provide digital health-saving technology to 9-1-1 call handlers and save lives by making accurate decisions by providing real-time assistance. This research aligns with the advancement of digital health technologies by exhibiting how NLP-driven decision support systems can revolutionize emergency healthcare, improve patient outcomes through real-time AI integration, and reduce errors.

This paper explores Text-to-Knowledge Graph (T2KG) construction, assessing Zero-Shot Prompting, Few-Shot Prompting, and Fine-Tuning methods with Large Language Models. Through comprehensive experimentation with Llama2, Mistral, and Starling, we highlight the strengths of FT, emphasize dataset size's role, and introduce nuanced evaluation metrics. Promising perspectives include synonym-aware metric refinement, and data augmentation with Large Language Models. The study contributes valuable insights to KG construction methodologies, setting the stage for further advancements.

Introduction: In today's competitive business landscape, organizations must enhance learning and adaptability to gain a strategic edge. While big data significantly influences organizational learning, a comprehensive tool to measure this capability has been lacking in the literature. This study aims to develop a valid and reliable scale to assess big data-based learning organization capability.

Methods: A two-phase research design was employed. In the first phase, Exploratory Factor Analysis (EFA) was conducted on data collected from 232 managers, identifying 22 items across three underlying factors. In the second phase, Confirmatory Factor Analysis (CFA) was applied to an independent sample (n = 128) to validate the scale's structure and its alignment with the theoretical model.

Results: The EFA results revealed a clear three-factor structure, and the CFA confirmed the model's fit to the data, demonstrating good psychometric properties. The final BD-LOC scale shows high internal consistency and construct validity.

Discussion: The BD-LOC scale provides organizations with a valuable tool to assess their big data-driven learning capabilities. It supports strategic decision-making, fosters innovation, and enhances operational efficiency. This study fills a significant gap in the literature and contributes to the effective implementation of digital transformation strategies in organizations.

Introduction: Long COVID, or Post-Acute Sequelae of SARS-CoV-2 infection (PASC), is a complex condition characterized by a wide range of persistent symptoms that can significantly impact an individual's quality of life and mental health. This study explores public perspectives on the mental health impact of Long COVID and the use of dietary supplements for recovery, drawing on social media content. It uniquely addresses how individuals with Long COVID discuss supplement use in the absence of public health recommendations.

Methods: The study employs the LISTEN method ("Collaborative and Digital Analysis of Big Qual Data in Time Sensitive Contexts"), an interdisciplinary approach that combines human insight and digital analysis software. Social media data related to Long COVID, mental health, and supplement use were collected using the Pulsar Platform. Data were analyzed using the free-text discourse analysis tool Infranodus and collaborative qualitative analysis methods.

Results: The findings reveal key themes, including the impact of Long COVID on mental health, occupational health, and the use of food supplements. Analysis of attitudes toward supplement use highlights the prevalence of negative emotions and experiences among Long COVID patients. The study also identifies the need for evidence-based recommendations and patient education regarding supplement use.

Discussion: The findings contribute to a better understanding of the complex nature of Long COVID and inform the development of comprehensive, patient-centered care strategies addressing both physical and mental health needs.

Chronic obstructive pulmonary disease (COPD), a major cause of global mortality, necessitates novel therapies targeting lung function and remodeling. Their effect on emphysema formation is initially investigated using mouse models by analyzing histological lung sections. The extent of airspace enlargement that is characteristic for emphysema is quantified by manual assessment of the mean linear intercept (MLI) across multiple histological microscopy images. Besides being tedious and cost intensive, this manual task lacks scientific comparability due to complexity and subjectivity. In order to continue with the well-established practice and to preserve the comparability of study results, we propose a deep learning-based approach for automating the determination of MLI in histological lung sections utilizing the AutoML software AIxCell which is specialized for the domain of semantic segmentation-based cell culture and tissue analysis. We develop and evaluate our image processing pipeline on stained histological microscope images that stem from a study including two groups of C57BL/6 mice where one group was exposed to cigarette smoke while the control group was not. The results indicate that the AIxCell segmentation algorithm achieves excellent performance, with IoU scores consistently exceeding 90%. Furthermore, the automated approach consistently yields higher MLI values compared to the manually generated values. However, the consistent nature of this discrepancy suggests that the automated approach can be reliably employed without any limitations. Moreover, it demonstrates statistical significance in distinguishing between smoker's and non-smoker's lungs.

This study presents a methodological framework for visualizing the alignment between complex research systems and the Sustainable Development Goals (SDGs), using CIRAD as a case study. By leveraging advanced data visualization and bibliometric analysis, the research maps CIRAD's publications to the SDGs and explores thematic priorities and institutional collaborations. The findings underscore CIRAD's significant contributions to climate action, food security, biodiversity conservation, and rural development. The integration of complex systems theory and network analysis enhances understanding of SDG interlinkages and provides actionable insights for strategic decision-making in research governance.