Search Results (194)

The deployment of Deep Learning (DL) for anomaly detection in Industrial IoT (IIoT) is critically hampered by the non-stationary nature of industrial data streams and the lack of forensic-grade explainability. This systematic review synthesizes 48 peer-reviewed studies (2021–2025) to quantify the performance collapse of static models under concept drift and to establish operational criteria distinguishing post hoc feature attribution (Type A XAI) from forensic root-cause diagnosis (Type B XAI). Our analysis reveals three critical findings: (1) static DL models suffer a 15–22% F1-score degradation across wastewater, manufacturing, and energy sectors when deployed in non-stationary environments, rendering them operationally non-viable without continuous adaptation; (2) the current literature remains saturated with Type A explainability (80% of corpus through 2023), creating a Forensic Gap where operators receive statistical correlations but lack actionable maintenance directives; and (3) emerging 2024–2025 research marks a paradigm shift toward Type B methodologies, yet no unified framework bridges real-time detection with deep causal reasoning. To address these gaps, we contribute the following: (1) a validated operational taxonomy (Cohen’s $\kappa =0.84$) with reproducible five-criterion rubric enabling forensic XAI classification; (2) the first quantitative synthesis of drift penalties in industrial deployments; and (3) a three-tier Edge-Cloud Forensic XAI architecture that achieves 70% communication payload reduction via compressed latent vectors while integrating tnGAN-based data imputation (handling 20–30% missing data) and physics-guided causal reasoning engines. Our framework decouples millisecond-level edge detection from 1–3 s cloud-based forensic diagnosis, ensuring both operational responsiveness and actionable industrial insight. We conclude that the future of safety-critical IIoT demands “Forensic-by-Design” architectures leveraging machine unlearning for drift adaptation and LLM-based natural language interfaces for operator-facing explanations, positioning Industry 5.0 to bridge the gap between algorithmic detection and human-centered decision support. Full article

Recent advancements in generative AI have improved process mining, making workflow analysis more accessible and scalable. However, large language models lack structured reasoning and fail to capture sequential dependencies in workflows, limiting their effectiveness. While retrieval-augmented generation (RAG) improves contextual knowledge integration, it does not enforce process execution constraints, leading to inconsistencies in workflow modeling. To address this, we propose GRAG4PM, a graph retrieval-augmented generation framework designed for process mining. GRAG4PM introduces three key innovations: a hierarchical graph architecture that captures multi-level process semantics, an adaptive pruning mechanism that refines workflow representations while preserving critical information, and a process-specific dual-indexing scheme that adapts existing hybrid retrieval to the constraints of workflow graphs. Experimental results demonstrate that GRAG4PM significantly improves workflow validation, anomaly detection, and predictive analysis, particularly in handling incomplete information. Across the three evaluated process-aware tasks, workflow-state validation, violation checking, and next-activity prediction, GRAG4PM achieves an average Macro-F1 score of 0.7259, outperforming Graph RAG (0.4394) and standard RAG (0.4036). Under missing-attribute settings, GRAG4PM also maintains the highest average Macro-F1 across masking conditions, confirming its robustness in incomplete data scenarios. By balancing structured workflow constraints with flexible AI-driven retrieval, GRAG4PM enables dynamic, interpretable, and adaptive process mining solutions and explores promising directions for future research on improving workflow solutions for AI. Full article

Technical debt (TD) poses a significant systemic risk in aerospace systems engineering, yet existing frameworks inadequately address debt irreversibility at hardware–software integration boundaries. Current detection approaches operate on structured code artifacts rather than the unstructured test and evaluation (T&E) documentation where integration debt often becomes visible. This paper presents the Technical Debt Management Framework (TDMF), a proof-of-concept architecture for identifying, quantifying, and prioritizing TD across the systems engineering lifecycle. The TDMF proposes an integrative architecture combining leading indicator (LI) monitoring with an AI detection module using large language model (LLM) analysis to surface debt indicators within unstructured aerospace documentation. The framework is grounded in a systematic review of 143 publications and illustrated through retrospective application to the Hubble Space Telescope and Mars Climate Orbiter failures, with an Evidence Traceability Matrix bounding historical claims against hindsight bias. An initial pilot evaluation of the ATLAS prototype—conducted on a single-program aerospace T&E documentation using GPT-4 with expert annotation—yielded a preliminary F1 score of 0.82 and an observed 45% reduction in median review time, providing initial evidence of computational feasibility within that scope. The framework is positioned as an early-stage design-science artifact at Technology Readiness Level 2–3. Prospective multi-program validation constitutes the required next study. This work contributes a proof-of-concept management architecture, a documented prompt engineering approach for TD classification, and a structured research agenda for empirical validation for TD classification in mission-critical systems engineering. Full article

Building details are often treated as technical externalities, subordinate to form, image and architectural narrative. Reading details as liminal spaces reverses that hierarchy. The joint concentrates transitions between the inside and outside, public and private, exposure and protection, and these transitions are constructed as intervals, experienced through thickness, reveal, edge condition, shadow, touch, and the small resistances that accompany crossing. The article develops its analysis through archival hand-drawn detail drawings from the Azrieli Architectural Archive. It defines building details as both technical assemblies and threshold devices, points where architecture becomes accountable to perception as well as to climate, labor, regulation, and everyday use. A semiotic reading of large-scale sheets shows how line weight, hatching, notation, and layout encode priorities, marking boundaries between what must be precisely resolved and what may remain adjustable. The archive is treated as a laboratory of “detail families,” recurring junction types such as windows, stairs, and envelope edges that reveal office-specific languages of joining. Two case studies, by the architects Ram Karmi and Arieh Sharon with Eldar Sharon, show how micro-variations in depth, overlap, and edge control tune thresholds, producing perceptual tipping points where comfort can shift into irritation, calm into unease, and openness into vulnerability. Although grounded in a local archive, the argument addresses a broader condition of contemporary practice: standardization and digital production chains can relocate authorship and responsibility away from the joint, precisely where buildings most affect everyday conduct. The paper proposes a liminal literacy of detailing as both a historiographic method and a design ethic aimed at making threshold decisions legible, contestable, and accountable in present-day workflows. Full article

The National Register of Historic Places (NR) is a structured artifact of meaning-making that encodes disciplinary values linking architectural and cultural significance to wealth and stylistic distinction. In doing so, it systematically underrepresents vernacular, working-class, and the built environments of racially and ethnically marginalized communities. This paper uses artificial intelligence (AI) to examine how that meaning is constructed. We analyze the preservation record across three scales: a national dataset of 100,117 NR listings (1966–2025), a state-level profile of Illinois’s 1997 NR listings, and a close analysis of Lake Forest, Illinois, a community whose exceptional concentration of NR-listed estate architecture makes it an ideal site for examining how preservation significance has been defined and what it excludes. Two parallel AI methods are applied to eighteen Lake Forest nomination documents and their associated photographs. Natural Language Processing (NLP) analyzes nomination text to trace how preservation professionals connect buildings to cultural value; blind AI image analysis examines the same properties to assess how a model trained on cultural imagery constructs visual meaning independently. NLP analysis reveals a corpus dominated by architectural description, with social history, landscape, and labor systematically underrepresented. The visual analysis confirms and amplifies the nomination record’s class-based assumptions while reproducing the same omissions regarding labor, diversity, and community context. These findings inform debates about AI’s potential to audit existing listings and support nominations for underrepresented property types, while showing that without deliberate corrective design and policy reform, such tools are as likely to replicate the preservation system’s inequities as to repair them. Full article

Large Language Models (LLMs) offer state-of-the-art performance in natural language understanding and generation tasks. However, the deployment of leading commercial models for specialized tasks, such as e-commerce, is often hindered by high computational costs, latency, and operational expenses. This paper investigates the viability of smaller, open-weight models as a resource-efficient alternative. We present a methodology for optimizing a one-billion-parameter Llama 3.2 model for multilingual e-commerce intent recognition. The model was fine-tuned using Quantized Low-Rank Adaptation (QLoRA) on a synthetically generated dataset designed to mimic real-world user queries. Subsequently, we applied post-training quantization techniques, creating GPU-optimized (GPTQ) and CPU-optimized (GGUF) versions. Our results demonstrate that the specialized 1B model achieves 98.8% accuracy, approaching the performance of the significantly larger GPT-4.1 model. A detailed performance analysis revealed critical, hardware-dependent trade-offs: while 4-bit GPTQ reduced VRAM usage by 41%, it paradoxically slowed inference by 82% on an older GPU architecture (NVIDIA T4) due to dequantization overhead. Conversely, GGUF formats on a CPU achieved a speedup of up to $4.3\times$ in inference throughput and up to a 72% reduction in RAM consumption compared to the FP16 baseline. We conclude that small, properly optimized open-weight models are not just a viable but a more suitable alternative for domain-specific applications, offering state-of-the-art accuracy at a fraction of the computational cost. Full article

Background: Maritime operations increasingly rely on integrated, secure, and resilient architectures, yet the associated body of knowledge remains fragmented across conceptual, operational, logical, methodological, and governance-oriented perspectives. Objective: Our aim is to systematically review the literature on maritime integrated systems architecture in order to identify dominant themes, methodological tendencies, enabling technologies, and unresolved research gaps. Eligibility criteria: Peer-reviewed studies published in English were included when they addressed integrated systems architecture, or closely related architectural approaches, in maritime or naval contexts. Studies centred exclusively on isolated components, non-maritime settings without clear architectural transferability, or insufficient technical or methodological detail were excluded. Information sources: ACM Digital Library, IEEE Xplore, SpringerLink, ScienceDirect, MDPI, and IMarEST. Searches were carried out between January and March 2025, with the final search update for all sources completed in March 2025. Methods: The review was conducted and reported in accordance with PRISMA 2020. Three reviewers independently screened titles, abstracts, and full texts. Two reviewers independently extracted data, assessed methodological limitations and risk of bias using a review-specific qualitative appraisal framework, and evaluated the risk of bias due to missing results at the synthesis level. Disagreements were resolved through discussion and consensus, with third-reviewer arbitration when necessary. The synthesis combined qualitative thematic analysis across eleven predefined analytical categories with descriptive bibliometric and thematic mapping procedures. Results: Of 300 identified records, 60 studies met the inclusion criteria. Across non-mutually exclusive analytical categories, the literature was concentrated in Integrated Systems Architecture (52 studies), Development Processes (42), and Conceptual Models (37), whereas Zachman-based Methodology (4) and Maturity Models (3) were only marginally represented. Three recurrent patterns were observed across the corpus: the central role of cybersecurity and risk governance in architectural design; the growing importance of information technology and operational technology convergence for resilient monitoring, coordination, and decision support; and the increasing use of model-based and model-driven approaches to address architectural complexity. Overall confidence in the principal synthesized findings was judged to be moderate. Limitations: The review was limited to six databases and English-language publications, and the included studies varied in reporting depth, methodological transparency, and degree of empirical validation. Conclusions: The review organizes the field into a multilevel taxonomy spanning conceptual and operational models, logical and layered views, development processes, reference architectures, enabling technologies, and maturity-related perspectives. Taken together, the findings suggest that research in this area has progressed more clearly in architectural representation and integration than in long-term evaluation, particularly with regard to the practical operationalization of Zachman-based approaches and the development of maritime-specific maturity assessment frameworks. Funding: This review received no external funding. Registration: The review was not prospectively registered, and no publicly accessible protocol was prepared. Full article

The growing adoption of Generative AI in education has created opportunities to automate complex pedagogical tasks, yet reliably and scalably assessing open-ended responses remains a challenge. This study proposes and evaluates an architectural solution for integrating a Large Language Model (LLM) into Moodle, combining Retrieval-Augmented Generation (RAG) and AI agent mechanisms to enable automated grading of open-ended student responses. A Moodle instance was deployed for experimental purposes, with 32 students across Bulgarian- and English-language sections, yielding data at the student (N = 32) and task (N = 160) levels, including AI-generated and instructor-assigned scores and system processing logs. The results demonstrate that the proposed system achieves substantial reductions in grading time while maintaining high agreement with expert assessments. Bias analysis revealed minimal systematic deviation across both language groups, indicating that the system preserves assessment objectivity without consistent over- or underestimation based on language. These findings suggest that a combined RAG and agentic LLM architecture can deliver efficient, accurate, and linguistically robust automated assessment within an LMS environment, offering practical design guidelines applicable to other educational platforms and similar systems. Full article

Research and development (R&D) organizations face significant operational bottlenecks due to the manual processing of diverse, unstructured documents. This paper presents the design, implementation, and pilot evaluation of an on-premise, multi-agent natural language processing (NLP) system developed for the GIG National Research Institute (GIG-NRI). Built upon a LangGraph architecture, the system utilizes open-weight large language models (LLMs) to perform zero-shot document classification, dynamic routing, and specialized information extraction. We rigorously evaluated the classification agent across twelve different local LLMs under two distinct testing regimes: first, using a strictly defined dataset of known administrative and scientific document types, and second, introducing a subset of out-of-distribution (unclassified) data to test real-world robustness. Our results demonstrate that the 70-billion parameter model (cogito:70b) achieved a peak accuracy of 97.3% in the first regime and maintained a strong 94.3% accuracy when confronted with out-of-spec data. However, our analysis reveals a critical operational trade-off regarding computational efficiency. The 24-billion parameter (magistral:24b) and 32-billion parameter (qwen3:32b) models emerged as the next best in overall accuracy while requiring less than half the processing time of their 70B counterpart. Notably, magistral:24b proved superior for strictly defined document streams, whereas qwen3:32b demonstrated greater robustness when handling out-of-distribution inputs. Furthermore, we demonstrate the efficacy of heterogeneous model assignments for complex multi-stage tasks, such as Scientific Article summarization via hierarchical Map-Reduce. Full article

Semantic communication (SemCom) has emerged as a task-oriented communication paradigm that prioritizes meaning delivery over exact bit recovery. The integration of generative artificial intelligence (GenAI) into SemCom further enables knowledge-guided inference, multimodal reconstruction, and semantic compression through architectures such as large language models, variational autoencoders, generative adversarial networks, and diffusion models. At the same time, this integration introduces new security and privacy risks, including semantic eavesdropping, model inversion, semantic jamming, covert backdoors, prompt manipulation, and knowledge-base leakage, which are not adequately captured by conventional communication security models. In this survey, we provide a security-centric review of GenAI-assisted semantic communication systems by organizing the literature according to threat models, attack surfaces, defence strategies, and semantic modalities across text, image, and multimodal settings. The survey was conducted using IEEE Xplore, ACM Digital Library, SpringerLink, arXiv, and Google Scholar. Approximately 180 papers were initially screened, and 53 representative studies published between 2021 and 2026 were selected for detailed review. Based on this analysis, we classify the major threats into adversarial perturbation, jamming, poisoning and backdoor attacks, privacy leakage and semantic eavesdropping, and generative-model-specific vulnerabilities involving diffusion, large language models, and multimodal foundation models. We further map the corresponding defences, including adversarial training, model ensembling, semantic-aware encryption, diffusion-guided denoising, privacy-preserving representation learning, and secure resource allocation. The survey also identifies persistent open challenges, including the lack of standardized semantic security metrics, unified benchmarks, cross-layer evaluation frameworks, and robust defences for GenAI-native and multimodal semantic communication systems. Overall, this work provides a structured reference for the design of secure, trustworthy, and attack-resilient generative semantic communication systems for future intelligent networks. Full article

The integration of Artificial Intelligence into educational systems has accelerated dramatically with the advent of Large Language Models (LLMs). However, two critical limitations constrain current AI-powered tutoring systems: LLMs hallucinate factually incorrect content in high-stakes pedagogical contexts, and existing systems lack standardized mechanisms to dynamically access and synthesize knowledge from heterogeneous educational sources, including learning management systems, open-access textbook repositories, assessment databases, and real-time educational APIs. This paper presents a systematic survey of the convergence of Retrieval-Augmented Generation (RAG) and the Model Context Protocol (MCP) in educational AI applications. Based on our taxonomy, we identify a critical architectural gap: no current system simultaneously achieves multi-source curriculum retrieval, standardized tool orchestration, learner-adaptive personalization, and citation-aware generation within a unified framework. To address this, we propose EduMSRA (Educational Multi-Source Research Agent)—a novel architecture comprising a Hierarchical Educational RAG Pipeline, an MCP-based Curriculum Tool Orchestration Layer, a Conflict-Aware Fusion Module (CAFM), a Learner Profile Manager (LPM), and a Pedagogical Policy Agent (PPA) aligned with Bloom’s taxonomy. We further provide a comprehensive experimental design road map specifying nine publicly available benchmark datasets and four evaluation experiments. Additionally, we conduct three Bayesian empirical analyses: (1) a random-effects meta-analysis of 12 RAG studies indicating a positive effect direction ($\widehat{\mu }=0.511$, 95% HDI: $[0.250,0.790]$), $I^2=99.3\%$ heterogeneity flagged as indicative), (2) a BKT simulation illustrating adaptive scaffolding dynamics across five learner profiles, and (3) a Beta-Binomial difficulty characterization of nine benchmark datasets. Our analysis demonstrates that EduMSRA offers a principled, scalable path toward adaptive, grounded, and pedagogically aligned AI tutoring agents. Full article

Urban radio spectrum monitoring is becoming increasingly complex due to the rapid growth of wireless devices, unauthorized emissions, and dynamic electromagnetic environments in smart cities. Traditional spectrum analysis approaches, based on manual operation or static detection techniques, are no longer sufficient to ensure scalable, autonomous, and secure monitoring. The convergence of two emergent technologies—Large Language Models (LLMs) and the Model Context Protocol (MCP)—facilitates a fundamental shift in radio monitoring. We define this as the AICEBERG paradigm: a novel, stratified architecture where a high-level, intelligent agentic interface (the peak) abstracts the underlying complexity of SCPI-driven hardware integration and radio governance protocols (the foundational base). This autonomous framework provides the necessary objective rigor to audit the stochastic ‘ocean of electromagnetic waves’ characteristic of modern smart cities, ensuring a stable platform for regulatory enforcement amidst high-density signal interference. The proposed system implements a three-layer processing flow, enabling high-level natural language commands to be translated into validated and secure hardware actions on RF spectrum analyzers. A dual-server design separates operational execution from safety validation, ensuring controlled SCPI command handling, parameter verification, and instrument health monitoring. Experimental validation demonstrates the feasibility of autonomous measurement execution. The results show that the proposed architecture reduces human dependency, enhances reproducibility and lowers the expertise barrier required for RF spectrum surveillance. To the best of our knowledge, AICEBERG represents one of the first integrated frameworks to bridge LLMs with SCPI-compliant hardware through the MCP for autonomous radio governance. Full article

Hate speech on social media reproduces norms of inequality and gender stereotypes, disproportionately affecting women. This study proposes a hybrid approach that integrates emotional tone classification with explicit hostility detection to strengthen preventive moderation. We constructed a corpus from three open data sets (1,236,371 records; 1,003,991 after ETL) and represented the text using TF-IDF and contextual RoBERTa embeddings. We trained individual models (RoBERTa fine-tuned, Random Forest, and XGBoost) and a stacking metamodel (Gradient Boosting) that combines their probabilities. On the test set, the ensemble outperformed the base classifiers, achieving accuracy of 0.93 in hate detection and 0.90 in emotion classification, with an AUC of 0.98 for emotion classification. We implemented a RESTful API and a web client to validate the moderation flow before publication, along with an administration panel for auditing. Performance tests in a prototype deployment (Google Colab exposed through an Ngrok tunnel) provided proof-of-concept validation, revealing concurrency limitations from around 300 users due to infrastructure constraints. In general, the results indicate that incorporating emotional tone analysis improves the model’s ability to identify implicit hostility and offers a practical way to promote safer digital environments. The probabilistic outputs produced by the ensemble model were subsequently analyzed using the Bayesian Calibration and Optimal Design under Asymmetric Risk (BACON-AR) framework, which serves as a mathematical post hoc decision layer for evaluating classification behaviour under unequal error costs. Rather than modifying the trained architecture or improving its predictive performance, the framework identifies a cost-sensitive operating threshold that minimizes the total expected risk under the selected asymmetric cost configuration. The experiments were conducted using an English-language data set; therefore, the findings of this study are limited to hate speech detection in English. Full article

Background/Objectives: Large language models (LLMs) such as ChatGPT are rapidly being integrated into healthcare for tasks ranging from clinical documentation to diagnostic support. Current ethical discussions focus predominantly on bias, privacy, and accuracy, leaving three critical governance questions unresolved: What kind of knowledge does an LLM output represent in clinical reasoning? When is a clinician’s or patient’s trust in that output justified? Who bears responsibility when an AI-informed decision leads to patient harm? This study proposes the Epistemic Authority–Trust–Responsibility (ETR) Architecture, a normative conceptual framework that addresses these three questions as an integrated governance challenge. Methods: The framework was developed through normative conceptual analysis—a method that constructs governance proposals by synthesising philosophical principles, ethical theories, and empirical evidence. The literature was identified through structured searches of PubMed, PhilPapers, and EUR-Lex (January 2020–March 2026), drawing on the philosophy of medical knowledge, the ethics of trust and testimony, and the moral philosophy of responsibility. Results: The ETR Architecture produces four outputs: (i) a four-tier classification system that distinguishes LLM outputs—from administrative drafts to clinical evidence claims—and matches each tier to appropriate verification requirements; (ii) the concept of the ‘epistemic placebo’, formally defined as a governance measure that creates a documented appearance of compliance while lacking at least one operative element of genuine oversight; (iii) a model specifying four conditions under which trust in healthcare AI is justified; (iv) four testable hypotheses with associated research designs connecting governance design to trust calibration and patient safety. Conclusions: The 2025–2027 regulatory transition period offers a critical window for shaping how healthcare institutions govern AI. We argue that deploying LLMs without explicitly classifying their outputs and building appropriate oversight risks allows governance norms to be set by technology vendors rather than by evidence-informed, patient-centred policy. Full article

As artificial intelligence (AI) systems become increasingly autonomous, scalable, and embedded in critical digital infrastructure, AI safety has emerged as a significant consideration for cybersecurity, system reliability, and institutional trust. Advances in large language models and agentic systems expand the threat surface to include misalignment, large-scale misuse, opaque decision-making, and cross-border risk propagation, while existing debates remain fragmented across technical, ethical, and geopolitical domains. This paper conducts a structured comparative analysis of AI safety perspectives from ten influential thinkers, examining them across five dimensions and reframing their insights through a cybersecurity lens spanning national governance, industry standards, and firm-level design. Building on this synthesis, the study proposes a layered control architecture that organizes technical safeguards, governance mechanisms, and human oversight into a defense-in-depth structure. The framework is conceptual and theory-building, intended to clarify system-level security reasoning and support future empirical refinement across diverse institutional contexts. Full article