The increasing adoption of blockchain technology across industries highlights the need for adaptable and scalable solutions that lower technical barriers to entry. While blockchain offers security, decentralization, and transparency, existing frameworks often require specialized expertise, limiting widespread adoption. This study proposes FlexBoardChain, a novel, flexible, and democratic blockchain framework that facilitates seamless integration into various applications. Our approach addresses key challenges in smart contract implementation, including interoperability, modularity, and usability, by abstracting complexities and offering an accessible API. Through a literature review, we identified the limitations of existing blockchain frameworks. We designed FlexBoardChain to enhance economic viability and security while enabling non-experts to configure and deploy blockchain-based solutions effectively. We implemented a use case in a community-centric buy-sell group to validate our framework, demonstrating its applicability in realworld scenarios. Our findings suggest that FlexBoardChain may facilitate and foster broader blockchain adoption by providing a scalable, user-friendly, and adaptable solution tailored to industry needs.
Traffic congestion and vehicular emissions remain critical challenges in urban mobility. While reinforcement learning (RL) has shown promise in adaptive traffic signal control, conventional models may inadvertently encourage private vehicle use by merely reducing delay. In this study, we present a Q-learning-based traffic signal control framework enhanced with a vehicle prioritization mechanism for public transport and emergency vehicles. Implemented using the Simulation of Urban Mobility (SUMO), our approach is evaluated on a four-arm intersection scenario. Compared to fixed-time control, the standard Q-learning model achieves an $80 \%$ reduction in average vehicle delay and over $80 \%$ decrease in $C O_{2}$ emissions. The prioritized Q-learning variant further improves delay and emissions metrics while providing preferential treatment to high-impact vehicle categories. Crucially, this prioritization strategy helps incentivize public transport usage, mitigating the risk of increased private car dependence that often follows general congestion reduction efforts. Our results demonstrate that integrating vehicle prioritization into RL-based traffic control supports both sustainability and modal shift goals in intelligent transportation systems.
Digital Product Passports (DPPs) are emerging as foundational tools for transparency, traceability, and sustainability in global supply chains. As regulatory initiatives such as the European Union’s Ecodesign for Sustainable Products Regulation (ESPR) gain momentum, there is increasing demand for technical solutions that support decentralised management and retrieval of product lifecycle data. This paper proposes a lightweight, extensible architecture centred on the DPP Protocol—a general-purpose communication mechanism for interoperable access to product information. The approach is particularly suited to fragmented or highly variable production contexts, such as fashion and other sectors characterised by short product lifecycles, complex supply networks, and limited digital infrastructure. We present two complementary implementations: the DPP Software, a server-side system for managing product data, and the DPP Browser, a stateless client for visualising and querying digital passports. Together, these components demonstrate the feasibility and versatility of the proposed protocol. The solution emphasises interoperability, recursive querying, and role-based access control, offering a foundation for scalable adoption across diverse industrial and regulatory settings.
Industry 5.0 redefines industrial automation by emphasizing human-centricity, sustainability, and resilience. Within this paradigm, the Operator Digital Twin (ODT) has recently emerged as a digital counterpart of the human worker, integrating biometric, contextual, and behavioral data to enable adaptive interactions with machines. However, fully exploiting the potential of ODTs introduces significant challenges, including the dynamic management of AI functionalities, privacy preservation, and context-aware deployment across heterogeneous computing devices. This paper proposes a dynamic, privacy-aware framework for managing AI capabilities within ODTs. The approach supports real-time adaptation of machine learning models based on the operator’s condition and the computational constraints of devices such as smartphones and embedded systems. By leveraging edge processing, the architecture minimizes the exposure of sensitive biometric data while ensuring reliable functionality and compliance with privacy regulations. The framework is validated in a prototypical industrial testbed using real ML models and heterogeneous hardware, demonstrating its effectiveness in enabling context-driven and secure AI orchestration in ODT-enabled industrial environments.
The escalating complexity and length of online privacy policies pose a substantial obstacle to user comprehension, thereby undermining informed consent. While manual annotation efforts have improved transparency, they are inherently limited in scalability. To address this challenge, we introduce COAT (Comprehensive Online Agreement Transparency), a novel framework for the automated analysis and risk scoring of privacy policies using Large Language Models (LLMs). Within the COAT system, this paper presents a comparative study evaluating several LLMs, including OpenAI’s GPT series and open-source models. Our methodology benchmarks LLM performance against humanannotated privacy risk scores using a set of specific policy clauses. This study validates the feasibility of using LLMs for scalable, automated privacy policy evaluation and highlights the performance disparities among current models. To foster further research and evaluation, the resulting dataset of LLM-generated scores is made publicly available.
Process mining in the construction industry relies heavily on high-quality event logs, yet existing approaches primarily use structured data sources, leaving unstructured textual reports largely unexploited. This paper proposes a novel two-stage methodology to extract and preprocess event logs from unstructured, natural language on-site construction reports. Initially, domain-customized natural language processing techniques combined with IBM’s multilingual granite text embeddings transform noisy, heterogeneous task descriptions into dense vector representations. Semantic clustering using cosine similarity then aggregates linguistically diverse but conceptually identical activities into unified clusters. Subsequently, a temporal consolidation module leverages the longitudinal occurrence of activities across monthly reports, incorporating spatial metadata to infer continuous task durations and transitions. Evaluation on 2,424 task instances demonstrates effective reduction of redundancy and robust grouping into approximately $102-111$ semantically coherent clusters. This approach addresses a critical knowledge gap in construction process mining by enabling incorporation of rich textual data, thus enhancing event log completeness and quality for downstream analysis. This advancement enables enhanced workflow tracking and delay detection in construction projects, broadening the scope of process mining beyond traditional reliance on structured data sources. Our work lays foundational steps for integrating unstructured textual data into construction process analytics, offering practical insights for improved project monitoring and decision-making.
Modern supply chains still rely on centralized procurement platforms; hence they struggle with inefficiencies, lack of transparency, and limited adaptability to disruptions and market changes. This paper proposes a decentralized e-bidding platform for Business-to-Business (B2B) procurement, integrating blockchain technology and artificial intelligence driven Intelligent Autonomous Agents (IAAs) to ensure trustworthiness and efficiency. Blockchain’s immutable ledger provides transparency and security, while IAAs, powered by large language models (LLMs), automate operational bid evaluation and support strategic supplier selection. Simulations of two supply chain scenarios with varying information sharing levels have demonstrated that higher transparency enhances efficiency. Results confirm the suitability of IAAs for real-time decision-making and blockchain for secure automation in supply chain applications.
The digital transformation of our society has deeply impacted the way we interact, communicate, and learn. This new digital era raises exceptional challenges for teaching and learning, particularly in adapting to individual learning paces, providing adequate progress traceability, and ensuring inclusive solutions for diverse learners. This paper introduces an Autonomic Cyber-Physical System for Education (A-CPS-E) that addresses these challenges through a framework combining teaching and learning design patterns with autonomous system capabilities. The A-CPS-E implements the autonomic computing paradigm to adapt learning experiences across four instructional interaction modes and generates Augmented Interactive Learning Objects (AILO) using AI support to enhance the teaching and learning process. We deployed the system within the Erasmus+ Connect Unita project, engaging 484 learners and 100 teachers across 12 universities in Europe. Resulting in the creation of 14 international training programs, including 65 learning paths and 400 learning objects, with $91.3 \%$ remote engagement. Preliminary results have allowed us to evaluate the benefits of the A-CPS-E in providing adaptive, traceable, and inclusive educational experiences that transcend traditional boundaries in higher education.
The Edge-Cloud Continuum (ECC) represents an emerging distributed, multi-tier computing infrastructure that integrates the capabilities of cloud and edge computing to offer a versatile environment capable of hosting applications with specific requirements, such as low latency and high computing capacity, which neither cloud nor edge computing can individually guarantee. However, ensuring Quality of Service (QoS) across the ECC remains a complex task, particularly with regard to networking parameters, due to the limited controllability and predictability of public network segments. This work introduces a QoS-aware mechanism for constructing logical overlay networks that aim to ensure required network performance with high probability, despite the inherent variability of public network environments. The proposed approach builds the overlay by organizing nodes into layers, each defined by a target level of network Qos (e.g., latency). The construction process consists of two key stages: first, collecting network metrics among the available nodes, and second, assigning nodes to layers based on the observed metrics. To validate the proposed approach, this work presents a Monte Carlo simulation campaign that assesses its effectiveness in constructing overlay networks capable of meeting, with high probability, the required QoS levels.
Accurate and interpretable water quality prediction is crucial for environmental monitoring and public health. This study evaluates six machine learning models—Random Forest, Long Short-Term Memory (LSTM), K-Nearest Neighbors (KNN), Linear Regression, Ridge Regression, and Support Vector Regression (SVR)—using real-world groundwater data from ARPAE. Model performance was assessed via Mean Absolute Error (MAE) and Mean Squared Error (MSE), while SHAP values were employed for feature-level interpretability. Results indicate that Random Forest outperforms all models in both accuracy and explainability, whereas SVR demonstrates poor predictive capability and lacks meaningful interpretability. The study highlights the trade-offs between predictive power and transparency, offering insights for selecting appropriate models in water quality monitoring systems.
The growing adoption of artificial intelligence in business processes is profoundly changing collaboration practices, knowledge management and value creation. This paper offers a critical review of recent literature on this topic, with particular attention to the context of collaborative enterprises. Through a comparative analysis of four recent contributions on this topic, this paper highlights that the effectiveness of artificial intelligence in the context of collaborative enterprises depends not only on the technology itself, but also on the integration within organizational environments characterized by advanced collaboration and knowledge management practices. The paper provides an interpretation of collaborative enterprises as socio-technical ecosystems, in which artificial intelligence becomes a force that promotes innovation, but only if it is embedded in a culture of co-design and continuous learning. The surveyed findings offer valuable insights for rethinking organizational models and promoting a more conscious and synergistic adoption of artificial intelligence technologies in collaborative enterprises.
This study examines how organizations manage tensions that arise during the implementation of Explainable Artificial Intelligence (XAI) in product innovation. While XAI has advanced technically, its impact on organizational routines, human interpretation of algorithmic outputs, and human-AI dynamics remains underexplored. Drawing on a qualitative case study of a global confectionery firm, we analyze the introduction of an XAI solution using participatory action research and paradox theory. We identify four persistent tensions: automation vs. human judgment, transparency vs. complexity, speed vs. accuracy, and standardization vs. customization. Rather than resolving these conflicts, the organization navigated them through both/and strategies that enabled human-AI collaboration. The findings extend paradox theory to XAI-driven innovation and contribute to digital transformation literature by showing how explainability supports knowledge articulation, learning, and adoption. The study also offers practical guidance for designing XAI systems that complement human expertise in complex innovation settings.
The analysis of aerial images to detect objects of interest is an active research field as its results can be very useful for planning crops, estimating gains and costs, etc. This paper proposes a novel way to distinguish among some categories of crops in a field, as well as parts representing man-made objects, such as roads and buildings. For this, advanced YOLO models were trained using an accurately labeled data set, then validated and tested to assess the accuracy and precision of the detection. The outcome of the best trained YOLO model shows high precision. Therefore, the trained model can be very useful in scanning areas and systematically detecting the state of crops. Moreover, the high-quality labeled image data set can be used as a reference set to test other approaches.
Fake news diffusion is a primary driver of misinformation. Analyzing deliberately false and misleading content is tough because social media platforms make it incredibly easy to create and spread huge amounts of information quickly. The intricate dynamics of fake news propagation demand the availability of ready-to-use frameworks for its analysis. This paper explores the automatic topic identification component of SPREADSHOT, a graph-based method designed to analyze fake news dissemination by examining two key factors: spreaders and topics. When it comes to news content, fake news frequently revolves around rapidly evolving topics due to its strong connection to current events. Consequently, topic modeling has gained significant traction for analyzing news articles. In our analysis, we explore two distinct topic modeling techniques: Latent Dirichlet Allocation (LDA) and BERTopic. While both offer valuable insights, we carefully justify which of these two techniques is best suited for integration into the SPREADSHOT framework for topic modeling.
This paper introduces an intelligent monitoring and anti-theft system designed for marine environments. Using Internet of Things technologies, the system utilises onboard devices that function as a finite-state machine, equipped with multiple sensors to gather speed, orientation and movement data. The onboard device is configured by the user with a personal device, which is used to arm and disarm the system via Bluetooth Low Energy. In case of theft, live position data is transmitted across a LoRa Mesh network, enabling long-range communication provided that enough boat nodes are available. The packets are transmitted using an adequate cryptography system based on AES128 keys. The data transmitted is recovered by a dock node, and it can be seen by the dock keeper and simultaneously provided to the users’ application.
Current health data management relies on centralized architectures that create a single point of failure, limit patient autonomy, and increase vulnerability to data breaches and vendor lock-in. This paper presents a decentralized approach to continuous health monitoring through the integration of wearable devices and distributed file systems. We implemented an Android application that collects physiological data and contextual information from a wearable device, storing it on the IPFS via Pinata API. Additionally, we propose a blockchain architecture for role-based access control. Performance evaluation comparing our IPFS-based implementation against Firebase Real-Time database reveals that the resource requirements remain negligible for modern smartphones while achieving significant benefits, including no single point of failure, enhanced data portability, and patient data sovereignty. The results demonstrate that decentralized health data management is technically feasible on mobile devices, offering an alternative approach to traditional centralized health data architectures.
Digital Twins (DTs) have emerged as a powerful paradigm for modeling, analyzing, and optimizing cyber-physical systems across a wide range of domains. Despite their growing adoption, the application of DTs to real-time systems—characterized by stringent timing constraints—remains a significant and underexplored area of research. Gaining a deeper understanding of the interaction between DTs and real-time systems is essential for extending the applicability of DT technologies to time-critical environments. This paper aims to establish a foundation for systematically addressing the challenges inherent in developing DTs for real-time systems. Specifically, we identify the key characteristics that a real-time system must possess to enable accurate and reliable DT replication. To illustrate these concepts, we present a case study in which a DT is employed to monitor and interact with a simulated robotic arm controlled by a real-time system. Our experimental findings highlight the strategic importance of this research direction and demonstrate the advantages of implementing a structured synchronization mechanism between the DT and its physical counterpart.
Financial crime is increasingly facilitated by technology and globalization, demanding advanced IT tools for detection. This paper presents an approach to automate the detection of suspicious activities in real time using Machine Learning (ML) techniques. The approach is designed and evaluated to operate under pragmatic operational constraints inherent to financial institutions, such as extreme class imbalance, stemming from the rarity of fraudulent events relative to legitimate transactions; evolving fraud patterns (concept drift), driven by adversarial adaptation; and significant delays in obtaining verified feedback. The paper aims to elucidate financial transaction monitoring in a data stream context and presents a developed streaming ML pipeline and an experimental testbed.
In this paper, we introduce a system designed to track the social interactions of individuals seated around a table, such as during work sessions or breaks. The system employs sensor technologies and data processing algorithms to monitor the seating arrangement and identify patterns of social engagement among participants. This represents an initial step towards developing a comprehensive system that provides deeper insights into collaborative behaviours and interpersonal connections within organizational settings.
Smart contracts, which are self-executing agreements, have a huge range of possible uses from finance to supply chain management. To avoid costly errors and vulnerabilities, it is crucial to guarantee the accuracy and reliability of these contracts. This paper explores the convergence of Blockchain technology, particularly Ethereum’s smart contracts, and Business Process Modeling (BPM), capitalizing on the synergies between these domains. We propose that viewing smart contracts as akin to business processes can significantly enhance the verification of Blockchain-based applications, addressing critical challenges in smart contract correctness and security. In this work we employ a formal verification approach based on Coloured Petri Nets and Linear Temporal Logic to detect potential vulnerabilities in Solidity smart contracts while considering their behavioral context as a business process model.
