Big Data and Cognitive Computing

This paper presents an in-production Rhetorical Role Labeling (RRL) classifier developed for Hungarian judicial decisions. RRL is a sequential classification problem in Natural Language Processing, aiming to assign functional roles (such as facts, arguments, decision, etc.) to every segment or sentence in a legal document. The study was conducted on a human-annotated sentence-level RRL corpus and compares multiple neural architectures, including BiLSTM, attention-based networks, and a support vector machine as baseline. It further investigates the impact of late chunking during vectorization, in contrast to classical approaches. Results from tests on the labeled dataset and annotator agreement statistics are reported, and performance is analyzed across architecture types and embedding strategies. Contrary to recent findings in retrieval tasks, late chunking does not show consistent improvements for sentence-level RRL, suggesting that contextualization through chunk embeddings may introduce noise rather than useful context in Hungarian legal judgments. The work also discusses the unique structure and labeling challenges of Hungarian cases compared to international datasets and provides empirical insights for future legal NLP research in non-English court decisions. Full article

This work represents the natural continuation of the development of the cognitive architecture developed and named Sophimatics, organically integrating the spatio-temporal processing mechanisms of the Super Time Cognitive Neural Network (STCNN) with the advanced principles of Sophimatics. Sophimatics’ goal is as challenging as it is fraught with obstacles, but its ultimate aim is to achieve a more humanized post-generative artificial intelligence, capable of understanding and analyzing context and evaluating the user’s purpose and intent, viewing time not only as a chronological sequence but also as an experiential continuum. The path to achieving this extremely ambitious goal has been made possible thanks to some previous work in which the philosophical thinking of interest in AI was first inherited as the inspiration for the aforementioned capabilities of the Sophimatic framework, then the issue of mapping concepts and philosophical thinking in Sophimatics’ AI infrastructure was addressed, and finally a cognitive-inspired network such as STCNN was created. This work, on the other hand, addresses the challenge of how to endow the infrastructure with both chronological and experiential time and its powerful implications, such as the innate ability to resolve paradoxes, which generative AI does not have among its prerogatives precisely because of structural limitations. To reach these results, the model operates in the two-dimensional complex time domain ℂ², extending cognitive processing capabilities through the implementation of dual temporal operators that simultaneously manage the real temporal dimension, where past, present, and future are managed and the imaginary one, that considers memory, creativity, and imagination. The resulting architecture demonstrates superior capabilities in resolving informational paradoxes and integrating apparently contradictory cognitive states, maintaining computational coherence through adaptive Sophimatic mechanisms. In conclusion, this work introduces Phase 4 of the Sophimatic framework, enabling management of two-dimensional time within a novel cognitively inspired neural architecture grounded in philosophical concepts. It connects with existing research on temporal cognition, hybrid symbolic–connectionist models, and ethical AI. The methodology translates philosophical insights into formal computational systems, culminating in a mathematical formalization that supports two-dimensional temporal reasoning and paradox resolution. Experimental results demonstrate efficiency, predictive accuracy, and computational feasibility, highlighting potential real-world applications, future research directions, and present limitations. Full article

Federated learning has gained popularity in recent years to enhance IoT security because the model allows decentralized devices to collaboratively learn a shared model without exchanging raw data. Despite its privacy advantages, federated learning is vulnerable to poisoning attacks, where malicious devices introduce manipulated data or model updates to corrupt the global model. These attacks can degrade the model’s performance or bias its outcomes, making it difficult to ensure the integrity of the learning process across decentralized devices. In this research, our goal is to develop a defense mechanism against poisoning attacks in federated learning models. Specifically, we develop a spacetime model, that combines the three dimensions of space and the one dimension of time into a four-dimensional manifold. Poisoning attacks have complex spatial and time relationships that present identifiable patterns in that manifold. We propose SpaceTime-Deep Similarity Defense (ST-DSD), a deep learning recurrent neural network that includes space and time perceptions to provide a defense against poisoning attacks for federated learning models. The proposed mechanism is built upon a time series regression many-to-one architecture using spacetime relationships to provide an adversarial trained deep learning poisoning defense. Simulation results show that SpaceTime defense outperforms existing solutions for poisoning defenses in IoT environments. Full article

In the context of remanufacturing, particularly mobile device refurbishing, effective operator training is crucial for accurate defect identification and process inspection efficiency. This study examines the application of Natural Language Processing (NLP) techniques to evaluate operator expertise based on subjective textual responses gathered during a defect analysis task. Operators were asked to describe screen defects using open-ended questions, and their responses were compared with expert responses to evaluate their accuracy and consistency. We employed four NLP models, including finetuned Sentence-BERT (SBERT), pre-trained SBERT, Word2Vec, and Dice similarity, to determine their effectiveness in interpreting short, domain-specific text. A novel disagreement tagging framework was introduced to supplement traditional similarity metrics with explainable insights. This framework identifies the root causes of model–human misalignment across four categories: defect type, severity, terminology, and location. Results show that a finetuned SBERT model significantly outperforms other models by achieving Pearsons’s correlation of 0.93 with MAE and RMSE scores of 0.07 and 0.12, respectively, providing more accurate and context-aware evaluations. In contrast, other models exhibit limitations in semantic understanding and consistency. The results highlight the importance of finetuning NLP models for domain-specific applications and demonstrate how qualitative tagging methods can enhance interpretability and model debugging. This combined approach indicates a scalable and transparent methodology for the evaluation of operator responses, supporting the development of more effective training programmes in industrial settings where remanufacturing and sustainability generally are a key performance metric. Full article

Massive Open Online Courses (MOOCs) have gained increasing popularity in recent years, highlighting the growing importance of effective course recommendation systems (CRS). However, the performance of existing CRS methods is often limited by data sparsity and suffers under cold-start scenarios. One promising solution is to leverage course-level conceptual information as side information to enhance recommendation performance. We propose a general framework for integrating LLM-generated concepts as side information into various classic recommendation algorithms. Our framework supports multiple integration strategies and is evaluated on two real-world MOOC datasets, with particular focus on the cold-start setting. The results show that incorporating LLM-generated concepts consistently improves recommendation quality across diverse models and datasets, demonstrating that automatically generated semantic information can serve as an effective, reusable, and scalable source of side knowledge for educational recommendations. This finding suggests that LLMs can function not merely as content generators but as practical data augmenters, offering a new direction for enhancing robustness and generalizability in course recommendation. Full article

Depression represents a critical global mental health challenge, with social media offering unprecedented opportunities for early detection through computational analysis. We propose a novel BERT–CNN–BiLSTM architecture with attention mechanisms that systematically integrate emoji usage patterns—fundamental components of digital emotional expression overlooked by existing approaches. Evaluated on the SuicidEmoji dataset, our model achieves 97.12% accuracy, 94.56% precision, 93.44% F1-score, 85.67% MCC, and 91.23% AUC-ROC. Analysis reveals distinct emoji patterns: depressed users favour negative emojis (😔 13.9%, 😢 12.8%, 💔 6.7%) while controls prefer positive expressions (😂 16.5%, 😊 11.0%, 😎 10.2%). The attention mechanism identifies key linguistic markers, including emotional indicators, personal pronouns, and emoji features, providing interpretable insights into depression-related language. Our findings suggest that the integration of emojis substantially improves optimal social media-based mental health detection systems. Full article

As Large Language Models (LLMs) are core components in Retrieval-Augmented Generation (RAG) systems for knowledge-intensive tasks, concerns regarding hallucinations, redundancy, and unverifiable outputs have intensified, particularly in high-stakes domains, such as e-government. This study proposes a modular, multi-pipeline framework for statement-level faithfulness evaluation for characterizing hallucination and redundancy across both simple and agentic RAG pipelines. Using GPT-4.1, Claude Sonnet-4.0, and Gemini 2.5 Pro as LLM-based judges, this study examines how tool-specific attribution within agentic multi-tool architectures influences the interpretability and traceability of the generated content. By using a modular agentic RAG framework combining symbolic (GraphRAG), semantic (embedding), and real-time (web) retrieval, we benchmark hallucination and redundancy patterns, using state-of-the-art LLM judges. The study examines RAG and agent-based pipelines that attribute outputs to distinct tools, in contrast to traditional single-source RAG systems that rely on aggregated retrieval. Using e-government data sourced from the European Commission’s Press Corner, our evaluation framework assesses not only the frequency, but also the source-aware detectability of hallucinated content. The findings provide actionable insights into how source granularity and retrieval orchestration impact faithfulness evaluation across different pipeline architectures, while also suggesting new directions for explainability-aware RAG design. The study contributes a reproducible, modular framework for automated faithfulness assessment, with implications for transparency, governance compliance, and trustworthy AI deployment. Full article

The increasing scale of modern datasets has created a significant computational bottleneck for traditional scientific and statistical algorithms. To address this problem, the current paper describes and validates a high-performance method based on adaptive spline interpolation that can dramatically accelerate the calculation of foundational scientific and statistical functions. This is accomplished by constructing parsimonious spline models that approximate their target functions within a predefined, highly precise maximum error tolerance. The efficacy of the adaptive spline-based solutions was evaluated through benchmarking experiments that compared spline models against the widely used algorithms in the Python SciPy library for the normal, Student’s t, and chi-squared cumulative distribution functions. Across 30 trials of 10 million computations each, the adaptive spline models consistently achieved a maximum absolute error of no more than 1 × 10⁻⁸ while simultaneously ranging between 7.5 and 87.4 times faster than their corresponding SciPy algorithms. All of these improvements in speed were observed to be statistically significant at p < 0.001. The findings establish that adaptive spline interpolation can be both highly accurate and much faster than traditional scientific and statistical algorithms, thereby offering a practical pathway to accelerate both the analysis of large datasets and the progress of scientific inquiry. Full article

As a vital carrier of human intangible culture, dance plays an important role in cultural transmission through digital generation. However, existing dance generation methods rely heavily on high-precision motion capture and manually annotated datasets, and they fail to effectively model the culturally distinctive movements of Chinese ethnic folk dance, resulting in semantic distortion and cross-modal mismatch. Building on the Chinese traditional ethnic Helou Dance, this paper proposes a culture-aware Chinese ethnic folk dance generation framework, CAFE-Dance, which dispenses with manual annotation and automatically generates dance sequences that achieve high cultural fidelity, precise music synchronization, and natural, fluent motion. To address the high cost and poor scalability of cultural annotation, we introduce a Zero-Manual-Label Cultural Data Construction Module (ZDCM) that performs self-supervised cultural learning from raw dance videos, using cross-modal semantic alignment and a knowledge-base-guided automatic annotation mechanism to construct a high-quality dataset of Chinese ethnic folk dance covering 108 classes of curated cultural attributes without any frame-level manual labels. To address the difficulty of modeling cultural semantics and the weak interpretability, we propose a Culture-Aware Attention Mechanism (CAAM) that incorporates cultural gating and co-attention to adaptively enhance culturally key movements. To address the challenge of aligning the music–motion–culture tri-modalities, we propose a Tri-Modal Alignment Network (TMA-Net) that achieves dynamic coupling and temporal synchronization of tri-modal semantics under weak supervision. Experimental results show that our framework improves Beat Alignment and Cultural Accuracy by 4.0–5.0 percentage points and over 30 percentage points, respectively, compared with the strongest baseline (Music2Dance), and it reveals an intrinsic coupling between cultural embedding density and motion stability. The code and the curated Helouwu dataset are publicly available. Full article

Pooling strategies are fundamental to convolutional neural networks, shaping the trade-off between accuracy, robustness to spatial variations, and computational efficiency in modern visual recognition systems. In this paper, we present and validate ECA110-Pooling, a novel rule-based pooling operator inspired by elementary cellular automata. We conduct a systematic comparative study, benchmarking ECA110-Pooling against conventional pooling methods (MaxPooling, AveragePooling, MedianPooling, MinPooling, KernelPooling) as well as state-of-the-art (SOTA) architectures. Experiments on three benchmark datasets—ImageNet (subset), CIFAR-10, and Fashion-MNIST—across training horizons ranging from 20 to 50,000 epochs show that ECA110-Pooling consistently achieves higher Top-1 accuracy, lower error rates, and stronger F1-scores than traditional pooling operators, while maintaining computational efficiency comparable to MaxPooling. Moreover, when compared with SOTA models, ECA110-Pooling delivers competitive accuracy with substantially fewer parameters and reduced training time. These results establish ECA110-Pooling as a principled and validated approach to image classification, bridging the gap between fixed pooling schemes and complex deep architectures. Its interpretable, rule-based design highlights both theoretical significance and practical applicability in contexts that demand a balance of accuracy, efficiency, and scalability. Full article

Following PRISMA-ScR guidelines, this scoping review systematically maps the landscape of Large Language Models (LLMs) in mechanical engineering. A search of four major databases (Scopus, IEEE Xplore, ACM Digital Library, Web of Science) and a rigorous screening process yielded 66 studies for final analysis. The findings reveal a nascent, rapidly accelerating field, with over 68% of publications from 2024 (representing a year-on-year growth of 150% from 2023 to 2024), and applications concentrated on front-end design processes like conceptual design and Computer-Aided Design (CAD) generation. The technological landscape is dominated by OpenAI’s GPT-4 variants. A persistent challenge identified is weak spatial and geometric reasoning, shifting the primary research bottleneck from traditional data scarcity to inherent model limitations. This, alongside reliability concerns, forms the main barrier to deeper integration into engineering workflows. A consensus on future directions points to the need for specialized datasets, multimodal inputs to ground models in engineering realities, and robust, engineering-specific benchmarks. This review concludes that LLMs are currently best positioned as powerful ‘co-pilots’ for engineers rather than autonomous designers, providing an evidence-based roadmap for researchers, practitioners, and educators. Full article

Driving safety education remains a critical societal priority, and understanding traffic rules is essential for reducing road accidents and improving driver awareness. This study presents the development and evaluation of a virtual simulator for learning traffic rules, incorporating spherical video technology and interactive training scenarios. The primary objective was to enhance the accessibility and effectiveness of traffic rule education by utilizing modern virtual reality approaches without the need for specialized equipment. A key research component is using Petri net-based models to study the simulator’s dynamic states, enabling the analysis and optimization of system behavior. The developed simulator employs large language models for the automated generation of educational content and test questions, supporting personalized learning experiences. Additionally, a model for determining the camera rotation angle was proposed, ensuring a realistic and immersive presentation of training scenarios within the simulator. The system’s cloud-based, modular software architecture and cross-platform algorithms ensure flexibility, scalability, and compatibility across devices. The simulator allows users to practice traffic rules in realistic road environments with the aid of spherical videos and receive immediate feedback through contextual prompts. The developed system stands out from existing traffic rule learning platforms by combining spherical video technology, large language model-based content generation, and cloud architecture to create a more interactive, adaptive, and realistic learning experience. The experimental results confirm the simulator’s high efficiency in improving users’ knowledge of traffic rules and practical decision-making skills. Full article

This study addresses the problem of automatic attack detection targeting Linux-based machines and web applications through the analysis of system logs, with a particular focus on reducing the computational requirements of existing solutions. The aim of the research is to develop and evaluate the effectiveness of machine learning models capable of classifying system events as benign or malicious, while also identifying the type of attack under resource-constrained conditions. The Linux-APT-Dataset-2024 was employed as the primary source of data. To mitigate the challenge of high computational complexity, model optimization techniques such as parameter quantization, knowledge distillation, and architectural simplifications were applied. Experimental results demonstrate that the proposed approaches significantly reduce computational overhead and hardware requirements while maintaining high classification accuracy. The findings highlight the potential of optimized machine learning algorithms for the development of practical early threat detection systems in Linux environments with limited resources, which is particularly relevant for deployment in IoT devices and edge computing systems. Full article

This paper explores the replication of the DotA Plus prediction system using decision tree algorithms. The study implements and evaluates Extra Trees Classifier, Random Forest Classifier, and Hist Gradient Boosting Classifier, along with their combined average, for predicting the outcome of Defense of the Ancients (DotA) 2 matches. Data was collected using the OpenDotA API and the Steam API, and various features such as game duration, tower and barracks states, net-worth, assists, last hits, gold, level, gold per minute, and experience per minute were extracted for model training. Additionally, hero and item win rate features, derived from Dotabuff data, were incorporated to enhance the models’ predictive accuracy. The models were trained on datasets with varying match durations, including segments for matches under 10 min, between 10 and 20 min, and over 20 min. The experimental results show that the Extra Trees Classifier consistently outperformed other individual models and performed comparably to the averaged models, achieving a peak performance of 98.6% test accuracy on matches longer than 20 min when using match duration segmentation and hero/item embeddings. The study highlights the effectiveness of decision tree-based methods for real-time match outcome prediction in DotA 2 and offers insights into feature importance. The combined average of Extra Trees Classifier, Random Forest Classifier, and Hist Gradient Boosting Classifier models provides a robust and reliable prediction of DotA 2 match outcomes, thus showing potential as a real-time prediction system. Full article

Current crime spatiotemporal prediction models are limited by the insufficient ability of POI data to represent the continuity and mixed-use nature of urban spatial functions. To address this, our study applies an urban region representation method based on building footprints and validates its effectiveness in improving the accuracy of crime spatiotemporal prediction. Specially, we first use the Region Dual Contrastive Learning algorithm to generate region representations as a region graph by integrating building footprints and POI data. Then, the region graph combined with crime data is input into crime prediction models to predict four crime types, including Burglary, Robbery, Felony Assault, and Grand Larceny. Finally, ablation experiments are conducted to quantify the contribution of building footprints to prediction improvement. The experimental results on New York City crime data indicate that (1) the region representations significantly improve deep learning model performance, with the most improved LSTM achieving average increases of 5.66% in Macro-F1 and 18.57% in Micro-F1, particularly benefiting baseline models with lower accuracy, and (2) the region representations yield more significant improvements for low-frequency crime categories and mitigates temporal memory decay in long-term predictions. These findings confirm that incorporating urban region representation based on building footprints effectively enhances crime spatiotemporal prediction performance, providing a more precise and efficient tool for urban security management to optimize police resource allocation and crime prevention strategies. Full article

To enable fully automated medicine warehousing in intelligent pharmacy systems, accurately detecting disordered, stacked pillboxes is essential. This paper proposes a high-precision detection algorithm for such scenarios based on an improved YOLOv8 framework. The proposed method integrates a novel convolutional module that replaces traditional stride convolutions and pooling layers, enhancing the detection of small, low-resolution targets in computer vision tasks. To further enhance detection accuracy, the Bi-Level Routing Attention (BiFormer) Vision Transformer is incorporated as a Cognitive Computing module. Additionally, the circular Smooth Label (CSL) technique is employed to mitigate boundary discontinuities and periodic anomalies in angle prediction, which often arise in the detection of rotated objects. The experimental results demonstrate that the proposed method achieves a precision of 94.24%, a recall of 90.39%, and a mean average precision (mAP) of 94.16%—improvements of 3.34%, 2.53%, and 3.35%, respectively, over the baseline YOLOv8 model. Moreover, the enhanced detection model outperforms existing rotated-object detection methods while maintaining real-time inference speed. To facilitate reproducibility and future benchmarking, the full dataset and source code used in this study have been released publicly. Although no standardized benchmark currently exists for pillbox detection, our self-constructed dataset reflects key industrial variations in pillbox size, orientation, and stacking, thereby providing a foundation for future cross-domain validation. Full article

High-performing medical Large Language Models (LLMs) typically require extensive fine-tuning with substantial computational resources, limiting accessibility for resource-constrained healthcare institutions. This study introduces a confidence-driven multi-model framework that leverages model diversity to enhance medical question answering without fine-tuning. Our framework employs a two-stage architecture: a confidence detection module assesses the primary model’s certainty, and an adaptive routing mechanism directs low-confidence queries to Helper models with complementary knowledge for collaborative reasoning. We evaluate our approach using Qwen3-30B-A3B-Instruct, Phi-4 14B, and Gemma 2 12B across three medical benchmarks; MedQA, MedMCQA, and PubMedQA. Results demonstrate that our framework achieves competitive performance, with particularly strong results in PubMedQA (0.95) and MedMCQA (0.78). Ablation studies confirm that confidence-aware routing combined with multi-model collaboration substantially outperforms single-model approaches and uniform reasoning strategies. This work establishes that strategic model collaboration offers a practical, computationally efficient pathway to improve medical AI systems, with significant implications for democratizing access to advanced medical AI in resource-limited settings. Full article

Social media platforms are currently confronted with a substantial problem concerning the presence of fake accounts, which pose a threat by spreading harmful content, spam, and misinformation. This study aims to address the problem by differentiating between fake and real X accounts (formerly Twitter). The need to mitigate the negative impact of fake accounts on online communities serves as the driving force for this work, with the goal of developing an effective method for identifying fake accounts and their fraudulent activities, such as posting harmful links, engaging in spamming behaviors, and disrupting online communities. The scope of this work focuses specifically on fake Twitter account detection. A comprehensive approach is taken, leveraging user information and tweets to discern between genuine and fake accounts. Various deep learning architectures are proposed and implemented, utilizing different optimizers and evaluating performance metrics. The models are trained and tested using a collected dataset, augmented to cover diverse real-life scenarios. The results show promising progress in distinguishing between fake and real accounts, revealing that the inclusion of tweet content along with user metadata does not significantly improve the classification of fake accounts. It also highlights the importance of selecting appropriate optimizers. The implications of this study are relevant to social media platforms, users, and researchers. The findings provide insights into combating fake accounts and their fraudulent activities, contributing to the enhancement of online community safety. While the research is specific to Twitter, the methodology and insights gained may be potentially generalizable to other social media platforms. Full article

Educational equity and access to quality learning opportunities represent fundamental pillars of sustainable societal development, directly aligned with the United Nations Sustainable Development Goal 4 (Quality Education). Student retention remains a critical challenge in higher education, with early disengagement strongly predicting eventual failure and limiting opportunities for social mobility. While machine learning models have demonstrated impressive predictive accuracy for identifying at-risk students, most systems prioritize performance metrics over practical deployment constraints, creating a gap between research demonstrations and real-world impact for social good. We present an accountable and interpretable decision support system that balances three competing objectives essential for responsible AI deployment: ultra-early prediction timing (day 14 of semester), manageable instructor workload (flagging 15% of students), and model transparency (multiple explanation mechanisms). Using the Open University Learning Analytics Dataset (OULAD) containing 22,437 students across seven modules, we develop predictive models from activity patterns, assessment performance, and demographics observable within two weeks. We compare threshold-based rules, logistic regression (interpretable linear modeling), and gradient boosting (ensemble modeling) using temporal validation where early course presentations train models tested on later cohorts. Results show gradient boosting achieves AUC (Area Under the ROC Curve, measuring discrimination ability) of 0.789 and average precision of 0.722, with logistic regression performing nearly identically (AUC 0.783, AP 0.713), revealing that linear modeling captures most predictive signal and makes interpretability essentially free. At our recommended threshold of 0.607, the predictive model flags 15% of students with 84% precision and 35% recall, creating actionable alert lists instructors can manage within normal teaching duties while maintaining accountability for false positives. Calibration analysis confirms that predicted probabilities match observed failure rates, ensuring trustworthy risk estimates. Feature importance modeling reveals that assessment completion and activity patterns dominate demographic factors, providing transparent evidence that behavioral engagement matters more than student background. We implement a complete decision support system generating instructor reports, explainable natural language justifications for each alert, and personalized intervention templates. Our contribution advances responsible AI for social good by demonstrating that interpretable predictive modeling can support equitable educational outcomes when designed with explicit attention to timing, workload, and transparency—core principles of accountable artificial intelligence. Full article