Unstructured Datasets Research Articles

Earthwork Network Architecture (ENA): Research for Earthwork Quantity Estimation Method Improvement with Large Language Model

Accurate earthwork quantity estimation is essential for effective project planning and cost management in the Architecture, Engineering, and Construction (AEC) industry. Traditional methods for quantity takeoff are often time-consuming and susceptible to human error, particularly when working with unstructured datasets such as CAD drawings. This study introduces the Earthwork Network Architecture (ENA), a novel deep learning framework that incorporates Large Language Models (LLMs), Multi-Layer Perceptron (MLP), Long Short-Term Memory (LSTM) networks, and Transformers to automate and enhance the accuracy of earthwork quantity estimation. We assume that if LLMs can be trained effectively using such unstructured construction dataset, the effects such as improved accuracy and the challenges of LLMs can be clearly examined. Among the architectures tested, the LLM-based ENA demonstrated superior performance, achieving faster convergence, greater loss reduction, and higher classification accuracy, with a Quantity Takeoff Classification accuracy of 97.17%. However, the LLMs required significantly more computational resources compared with other models. These findings suggest that LLMs, typically used in natural language processing, can be effectively adapted for complex AEC datasets. This study lays the groundwork for future AI-driven solutions in the AEC industry, underscoring the potential of LLMs and Transformers to automate the quantity takeoff process and manage multimodal data in construction projects.

Generative AI for Consumer Behavior Prediction: Techniques and Applications

Generative AI techniques, such as Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs), and transformers, have revolutionized consumer behavior prediction by enabling the synthesis of realistic data and extracting meaningful insights from large, unstructured datasets. However, despite their potential, the effectiveness of these models in practical applications remains inadequately addressed in the existing literature. This study aims to investigate how generative AI models can effectively enhance consumer behavior prediction and their implications for real-world applications in marketing and customer engagement. By systematically reviewing 31 studies focused on these models in e-commerce, energy data modeling, and public health, we identify their contributions to improving personalized marketing, inventory management, and customer retention. Specifically, transformer models excel at processing complicated sequential data for real-time consumer insights, while GANs and VAEs are effective in generating realistic data and predicting customer behaviors such as churn and purchasing intent. Additionally, this review highlights significant challenges, including data privacy concerns, the integration of computing resources, and the limited applicability of these models in real-world scenarios.

Parameterization before Meta-Analysis: Cross-Modal Embedding Clustering for Forest Ecology Question-Answering

In the field of forestry ecology, image data capture factual information, while literature is rich with expert knowledge. The corpus within the literature can provide expert-level annotations for images, and the visual information within images naturally serves as a clustering center for the textual corpus. However, both image data and literature represent large and rapidly growing, unstructured datasets of heterogeneous modalities. To address this challenge, we propose cross-modal embedding clustering, a method that parameterizes these datasets using a deep learning model with relatively few annotated samples. This approach offers a means to retrieve relevant factual information and expert knowledge from the database of images and literature through a question-answering mechanism. Specifically, we align images and literature across modalities using a pair of encoders, followed by cross-modal information fusion, and feed these data into an autoregressive generative language model for question-answering with user feedback. Experiments demonstrate that this cross-modal clustering method enhances the performance of image recognition, cross-modal retrieval, and cross-modal question-answering models. Our method achieves superior performance on standardized tasks in public datasets for image recognition, cross-modal retrieval, and cross-modal question-answering, notably achieving a 21.94% improvement in performance on the cross-modal question-answering task of the ScienceQA dataset, thereby validating the efficacy of our approach. Essentially, our method targets cross-modal information fusion, combining perspectives from multiple tasks and utilizing cross-modal representation clustering of images and text. This approach effectively addresses the interdisciplinary complexity of forestry ecology literature and the parameterization of unstructured heterogeneous data encapsulating species diversity in conservation images. Building on this foundation, intelligent methods are employed to leverage large-scale data, providing an intelligent research assistant tool for conducting forestry ecological studies on larger temporal and spatial scales.

XR-based interactive visualization platform for real-time exploring dynamic earth science data

The transition from 2D planar displays to immersive holographic 3D environments has brought advancements in visualization technology. However, there remains a lack of effective interactive visualization tools for complex multi-dimensional structured or unstructured datasets in immersive space. To address this gap, we have developed MetIVA, a state-of-the-art multiscale interactive data visualization platform that leverages the Extended Reality (XR) and cloud rendering technology for immersive data exploration. In this paper, we firstly outline the historical development of scientific visualization and the recent shift towards 3D and higher-dimensional visualization, and then basically introduce the conceptual framework and platform structure of MetIVA, and finally present the evaluation results from recruited potential users. The results confirm that MetIVA is a powerful tool to accelerate data exploration and decision-making processes. Its interactive and intuitive features, along with ongoing optimization efforts, make it a valuable tool for researchers and practitioners in the field of Earth science.

Dimension Reduction of Multidimensional Structured and Unstructured Datasets through Ensemble Learning of Neural Embeddings

Dimension reduction aims to project a high‐dimensional dataset into a low‐dimensional space. It tries to preserve the topological relationships among the original data points and/or induce clusters. NetDRm, an online dimensionality reduction method based on neural ensemble learning that integrates different dimension reduction methods in a synergistic way, is introduced. NetDRm is designed for datasets of multidimensional points that can be either structured (e.g., images) or unstructured (e.g., point clouds, tabular data). It starts by training a collection of deep residual encoders that learn the embeddings induced by multiple dimension reduction methods applied to the input dataset. Subsequently, a dense neural network integrates the generated encoders by emphasizing topological preservation or cluster induction. Experiments conducted on widely used multidimensional datasets (point‐cloud manifolds, image datasets, tabular record datasets) show that the proposed method yields better results in terms of topological preservation ( curves), cluster induction (V measure), and classification accuracy than the most relevant dimension reduction methods.

Explicit Incorporation of Spatial Autocorrelation in 3D Deep Learning for Geospatial Object Detection

Three-dimensional (3D) geospatial object detection has become essential for 3D geospatial studies driven by explosive growth in 3D data. It is extremely labor- and cost-intensive, though, as it often requires manual detection. Deep learning has been recently used to automate object detection within 3D context. Yet, addressing spatial dependency in 3D data and how it might inform deep learning for 3D geospatial object detection remains a significant challenge. Traditional models focus on the use of spatial properties, often overlooking color and contextual information. Exploiting these nonspatial attributes for 3D geospatial object detection thus becomes crucial. Our study pioneers explicit incorporation of spatial autocorrelation of color information into 3D deep learning for object detection. We introduce an innovative framework to estimate spatial autocorrelation, addressing challenges in unstructured 3D data sets. Our experiments suggest the effectiveness of incorporating spatial autocorrelation features in enhancing the accuracy of 3D deep learning models for geospatial object detection. We further investigate the uncertainty of such contextual information brought by diverse configurations, exemplified by the number of nearest neighbors. This study advances 3D geospatial object detection via using spatial autocorrelation to inform deep learning algorithms, strengthening the connection between GIScience and artificial intelligence and, thus, holding implications for diverse GeoAI applications.

Logistical and preference bias in participatory science butterfly data

The volume of and interest in unstructured participatory science data has increased dramatically in recent years. However, unstructured participatory science data contain taxonomic biases—encounters with some species are more likely to be reported than encounters with others. Taxonomic biases are driven by human preferences for different species and by logistical factors that make observing certain species challenging. We investigated taxonomic bias in reports of butterflies by characterizing differences between a dedicated participatory semi‐structured dataset, eButterfly, and a popular unstructured dataset, iNaturalist, in spatiotemporally explicit models. Across 194 butterfly species, we found that 53 species were overreported and 34 species were underreported in opportunistic data. Ease of identification and feature diversity were significantly associated with overreporting in opportunistic sampling, and strong patterns in overreporting by family were also detected. Quantifying taxonomic biases not only helps us understand how humans engage with nature but also is necessary to generate robust inference from unstructured participatory data.

MoConVQ: Unified Physics-Based Motion Control via Scalable Discrete Representations

In this work, we present MoConVQ, a novel unified framework for physics-based motion control leveraging scalable discrete representations. Building upon vector quantized variational autoencoders (VQ-VAE) and model-based reinforcement learning, our approach effectively learns motion embeddings from a large, unstructured dataset spanning tens of hours of motion examples. The resultant motion representation not only captures diverse motion skills but also offers a robust and intuitive interface for various applications. We demonstrate the versatility of MoConVQ through several applications: universal tracking control from various motion sources, interactive character control with latent motion representations using supervised learning, physics-based motion generation from natural language descriptions using the GPT framework, and, most interestingly, seamless integration with large language models (LLMs) with in-context learning to tackle complex and abstract tasks.

Brief Review and Primer of Key Terminology for Artificial Intelligence and Machine Learning in Hypertension.

Recent breakthroughs in artificial intelligence (AI) have caught the attention of many fields, including health care. The vision for AI is that a computer model can process information and provide output that is indistinguishable from that of a human and, in specific repetitive tasks, outperform a human's capability. The 2 critical underlying technologies in AI are used for supervised and unsupervised machine learning. Machine learning uses neural networks and deep learning modeled after the human brain from structured or unstructured data sets to learn, make decisions, and continuously improve the model. Natural language processing, used for supervised learning, is understanding, interpreting, and generating information using human language in chatbots and generative and conversational AI. These breakthroughs result from increased computing power and access to large data sets, setting the stage for releasing large language models, such as ChatGPT and others, and new imaging models using computer vision. Hypertension management involves using blood pressure and other biometric data from connected devices and generative AI to communicate with patients and health care professionals. AI can potentially improve hypertension diagnosis and treatment through remote patient monitoring and digital therapeutics.

Comparative Analysis and Mining of Unstructured Datasets from an Empirical Standpoint

Comparative Analysis and Mining of Unstructured Datasets from an Empirical Standpoint

Survey on Systematic Analysis of Deep Learning Models Compare to Machine Learning

This survey provides a comprehensive analysis of the systematic differences and advancements between deep learning (DL) and traditional machine learning (ML) models. By examining a wide array of research papers, the study highlights the unique strengths and applications of both methodologies. Deep learning, with its multi-layered neural networks, excels in handling large, unstructured datasets, making significant strides in image and speech recognition, natural language processing, and complex pattern recognition tasks. Conversely, traditional machine learning models, which rely on feature extraction and simpler algorithms, remain highly effective in structured data scenarios such as classification, regression, and clustering problems. The survey elucidates the criteria for choosing between DL and ML, focusing on factors like data size, computational resources, and specific application requirements. Furthermore, it discusses the evolving landscape of hybrid models that integrate DL and ML techniques to leverage the strengths of both approaches. This analysis provides valuable insights for researchers and practitioners aiming to deploy the most suitable AI models for their specific needs, emphasizing the importance of contextual understanding in the rapidly advancing field of artificial intelligence.

Making sense of student feedback and engagement using artificial intelligence

Making sense of student feedback and engagement is important for informing pedagogical decision-making and broader strategies related to student retention and success in higher education courses. Although learning analytics and other strategies are employed within courses to understand student engagement, the interpretation of data for larger data sets is more challenging and rarely pursued. This is concerning as data offers the potential for critical insights into engagement behaviour and the value students place on engagement. Artificial intelligence (AI) offers a revolutionary ability to make sense of data, with capacity for prediction and classification, by consuming vast amounts of structured and unstructured data sets. This paper reports on how AI methodologies (specifically, deep learning and natural language processing) were used to leverage labelled student feedback in terms of online engagement in five courses in a regional Australian university. This paper reinforces the value of AI as a viable and scalable multilayered analysis tool for analysing and interpreting student feedback, particularly for categorising student responses as to the types of engagement that they most valued to support their learning. The paper concludes with a discussion of suggested further refinement, including how the AI-derived data may add insights for informing pedagogical practice. Implications for practice or policy: AI offers an ability to make sense of large data sets in higher education courses. Teachers can use student feedback data categorised into types of engagement by AI to support reflection on what students value in their courses. Educators and key stakeholders can use the insights AI analysed data offers for informing pedagogical practice and decision-making in higher education to enhance student experiences.

Patient satisfaction with cancer care delivery through telemedicine at a comprehensive cancer center across three years during and after the COVID-19 pandemic.

1632 Background: Telemedicine, once sparingly utilized, surged during the COVID-19 pandemic to enhance care accessibility and safety concerns. Despite initial satisfaction, its usage remains highly variable across institutions. We sought to explore patient satisfaction at a comprehensive cancer center, associations with demographic characteristics, and change over time. Methods: A survey was sent to all first-time telemedicine users at Memorial Sloan Kettering Cancer Center from September 2021 to October 2023. Adult patients who had their initial telemedicine visit, regardless of specialty, received the questionnaire. It comprised 10 general questions about the telemedicine experience and 11 questions regarding clinical trial education and informed consent. To analyze the unstructured free text responses, we employed BERTopic, a topic modeling algorithm that leverages transformer-based embeddings to facilitate the creation of easily interpretable topics within a collection of text documents. Through topic modeling with BERT embeddings, we captured subtle nuances and semantic relationships in language to identify topics more effectively. Results: A total of 15,111 patients completed 15,251 questionnaires; 140 patients completing two surveys. Patient demographics are summarized in Table 1; the median age at encounter was 66 years and 98% of patients preferred English. Satisfaction levels were high, with 92% expressing willingness for future telemedicine visits and 90% recommending it to other patients. Notably, 21% found telemedicine superior to in-person visits, while 53% considered them equal. Further, satisfaction remained consistent across six-month intervals from 2021-2023. In 1606 (11%) surveys, patients noted discussing clinical trials; 17% preferred in-person discussions for the informed consent process while most either preferred telemedicine or had no preference. Topic modeling revealed 18 topics within the unstructured dataset; key topics included comments on video & audio quality, connectivity and technology issues, convenience for patients who live outside of the city and those with physical limitations, and long wait time to see a doctor. Conclusions: Our analysis confirms high patient satisfaction with telemedicine, including discussions on clinical trial enrollment, that has remained consistent in the years following the COVID-19 pandemic. Given variability in use, it is important to identify how to better deliver telemedicine to those patients who prefer it.[Table: see text]

Customising generative AI: Harnessing document retrieval and fine-tuning alternatives for dynamic marketing insights

This study delves into the transformative impact of leveraging large language models (LLMs) in marketing analytics, particularly emphasising a paradigm shift from fine-tuning models to the strategic application of document retrieval techniques and more. Focusing on innovative methods, such as retrieval augmented generation and low-rank adaptation, the paper explores how marketers can now activate against vast and unstructured datasets, such as call centre transcripts, unlocking valuable insights that were previously overlooked. By harnessing the power of document retrieval and adaptation, marketers can bring their data to life, enabling a more nuanced and adaptive approach to understanding consumer behaviour and preferences. This research contributes to the evolving landscape of applied marketing analytics by demonstrating the efficacy of document retrieval in enhancing the utilisation of LLMs for dynamic and data-driven marketing strategies.

An expert system for privacy-driven vessel detection harnessing YOLOv8 and strengthened by SHA-256

The integration of deep learning with security is undergoing rapid and substantial growth across numerous fields. This combination has led to propelled advancements in securing maritime surveillance data. Applications utilizing deep learning require large amounts of data to deliver precise results. However, Deep learning applications face significant issues like low accuracy, high computational complexity, and GPU utilization. Moreover, security features such as integrity and authenticity are notably absent. Hence, in this paper privacy preserved deep learning-based vessel monitoring system is developed which includes evaluation of integrity and authenticity-based surveillance. This model ensures accuracy without compromising on one way authenticity while also optimizing GPU utilization. Therefore, the YOLOv8 model is integrated with SHA-256 for tracking and classification, also ensuring data integrity and authentication of vessel data. The model is fed with class-balanced, unstructured dataset comprising 693 photo-realistic video sequences. It has three phases which are used for feature extraction and bounding box prediction and also includes CSPDarkNet53 as backbone, Spatial Pyramid Pooling (SPP) as neck layer and detection is done using head layer. They employ a 3 × 3 convolution for better feature extraction. The proposed model provides better performance than other state-of-the-art methods. Specifically, YOLOv8 achieves a 9.3 % increase in precision over YOLOv7.

A Framework for Enhancing Behavioral Science Research with Human-Guided Language Models

Many behavioral science studies result in large amounts of unstructured data sets that are costly to code and analyze, requiring multiple reviewers to agree on systematically chosen concepts and themes to categorize responses. Large language models (LLMs) have potential to support this work, demonstrating capabilities for categorizing, summarizing, and otherwise organizing unstructured data. In this paper, we consider that although LLMs have the potential to save time and resources performing coding on qualitative data, the implications for behavioral science research are not yet well understood. Model bias and inaccuracies, reliability, and lack of domain knowledge all necessitate continued human guidance. New methods and interfaces must be developed to enable behavioral science researchers to efficiently and systematically categorize unstructured data together with LLMs. We propose a framework for incorporating human feedback into an annotation workflow, leveraging interactive machine learning to provide oversight while improving a language model's predictions over time.

Shortcut learning in medical AI hinders generalization: method for estimating AI model generalization without external data

Healthcare datasets are becoming larger and more complex, necessitating the development of accurate and generalizable AI models for medical applications. Unstructured datasets, including medical imaging, electrocardiograms, and natural language data, are gaining attention with advancements in deep convolutional neural networks and large language models. However, estimating the generalizability of these models to new healthcare settings without extensive validation on external data remains challenging. In experiments across 13 datasets including X-rays, CTs, ECGs, clinical discharge summaries, and lung auscultation data, our results demonstrate that model performance is frequently overestimated by up to 20% on average due to shortcut learning of hidden data acquisition biases (DAB). Shortcut learning refers to a phenomenon in which an AI model learns to solve a task based on spurious correlations present in the data as opposed to features directly related to the task itself. We propose an open source, bias-corrected external accuracy estimate, PEst, that better estimates external accuracy to within 4% on average by measuring and calibrating for DAB-induced shortcut learning.

Bridging the innovation gap: Why organizational climate matters for leveraging innovation from supply networks

AbstractRecent studies have provided empirical evidence that innovation performance is related to the way a firm is embedded in its supply network, specifically the centrality of its network position, but it remains unclear why some firms can use inputs from suppliers better than others, despite having comparable structural characteristics in their supply networks. Drawing on theories of social networks and organizational climate, this study examines the role of buying firms' organizational climate for innovation. It uses several structured and unstructured datasets for S&P 500 firms and applies count regression models to test hypotheses. Supply network data from FactSet were analyzed to determine the degree centrality of a buying firm. Computer‐aided content analysis was used to capture the organizational climate of buying firms based on online employee reviews collected from Glassdoor. The results suggest a positive relationship between the degree centrality and the innovation performance of buying firms. Moreover, certain facets of the organizational climate related to learning, including rewards and career progress, as well as work pressure management, affect the link between the degree centrality of a buying firm and its innovation performance. In conclusion, this study enhances the understanding of the connection between supply networks and innovation. It highlights the crucial role of a firm‐level factor, specifically the influential facets of organizational climate for learning, in determining innovation performance.

An efficient fake news classification model based on ensemble deep learning techniques

The availability and expansion of social media has made it difficult to distinguish between fake and real news. Information falsification has exponentially increased as a result of how simple it is to spread information through sharing. Social media dependability is also under jeopardy due to the extensive dissemination of false information. Therefore, it has become a research problem to automatically validate information, specifically source, content, and publisher, to identify it as true or false. Despite its limitations, machine learning (ML) has been crucial in the categorization of information. Previous studies suggested three-step methods for categorising false information on social media. In the first step of the process, the data set is subjected to a number of pre-processing processes in order to transform unstructured data sets into structured data sets. The unknowable properties of fake news and the features are extracted by the Lexicon Model in the second stage. In the third stage of this research project, a feature selection method by WOA (Whale Optimization Algorithm) for weight value to tune the classification part. Finally, a Hybrid Classification model that is hybrid with a fuzzy based Convolutional Neural Network and kernel based support vector machine is constructed in order to identify the data pertaining to bogus news. However using single classifier for fake news detection produces the insufficient accuracy. To overcome this issue in this work introduced an improved model for fake news classification. To turn unstructured data sets into structured data sets, a variety of pre-processing operations are used on the data set in the initial phase of the procedure. The unknowable properties of fake news and the features are extracted by the Lexicon Model in the second stage. In the third stage of this research project, a feature selection method by COA (Coati Optimization Algorithm) for weight value to tune the classification part. Finally, an ensemble of RNN (Recurrent Neural Networks), VGG-16 and ResNet50.A classification model was developed to recognise bogus news information. Evaluate each fake news analysis' performance in terms of accuracy, precision, recall, and F1 score. The suggested model, out of all the methodologies taken into consideration in this study, provides the highest outcomes, according to experimental findings

Fast Disaster Event Detection from Social Media: An Active Learning Method

This study introduces a novel framework for fast disaster event detection from social media,hich incorporates active learning into Bi-directional Long Short-Term Memory (Bi-LSTM). In theace of increasing disaster-related information on social media platforms, our method addresses thehallenge of efficiently processing vast, unstructured datasets to accurately extract crucial disasterrelatednsights. Leveraging the contextual processing strengths of Bi-LSTM and the data efficiencyf active learning, our approach significantly outperforms several baseline models in accuracy,recision, recall, and F1-score. We demonstrate that our method effectively balances these metrics,nsuring reliable disaster detection while minimizing false positives and negatives. The studylso explores the impact of various model parameters on performance, offering insights for futureptimizations. Despite its promising results, the study acknowledges limitations such as datauality, representativeness, and computational resource requirements. Future work will focus onnhancing data preprocessing, expanding language and scenario applicability, and integrating theodel with additional technologies for broader disaster management applications.