Abstract
The increasing adoption of the Digital Twin concept and technology for managing complex physical assets has led to the emergence of Digital Twin Ecosystems, where interconnected digital twins generate additional value. However, ensuring seamless data sharing and interoperability among diverse systems presents significant challenges. Although research on digital twin architectures has advanced, gaps remain in addressing data governance, security, and stakeholders' trust. This study performs a comprehensive literature review to investigate architectural solutions to overcome challenges in digital twin ecosystems. The findings identify key requirements such as interoperability, governance, and data management, emphasizing the role of Data Spaces as enablers of secure data sharing. By structuring the requirements for digital twin ecosystem architectures, this paper identifies gaps suggesting future research on scalable and sustainable digital twin ecosystem implementations. These insights are expected to contribute to the development of frameworks that integrate technical advances with organizational and regulatory considerations, ultimately fostering the adoption of digital twin ecosystems across industries.

Abstract
The notion of Union is strength is essential for preserving public goods and mitigating public bads such as air quality. International environmental agreements serve this role by forming stable coalitions, in which agents join or leave based on free-riding incentives. Building on the Baliga-Maskin model, we show that such coalitions can emerge from a simple Markov chain mechanism where agents enter or exit through utility-based bargaining. However, stable coalition formation is challenging, as members may receive substantially lower utility than free-riders. This asymmetry gives rise to Barrett's paradox of cooperation: even with large coalitions and strong preferences among free-riders, overall utility may remain far below that of the grand coalition. Encouragingly, the paradox of cooperation can be resolved when free-riders have sufficiently low preferences.

Abstract
This paper evaluates an Augmented Reality (AR) solution designed to support quality control in a assembly line inspection station before body marriage at a European automotive manufacturer. A threephase methodology was applied: an AS-IS assessment, a formative evaluation of an intermediate prototype, and a summative evaluation under real production conditions. The AR solution aimed to improve task standardization, non-value-added time (NVAT), and enhance operator accuracy. The results showed that operators successfully developed inspections using the AR tool, identifying and correcting non-conformities (NOKs) while maintaining task duration. Participants valued having contextual information directly in their field of vision and reported increased rigor and consistency. However, usability and ergonomic improvements were noted, such as headset weight, gesture interaction, and visibility over dark components. The findings highlight AR's potential to support operator autonomy and accuracy in industrial environments while emphasizing the need for human-centered design and integration to ensure long-term adoption.

Abstract

Abstract
State-of-the-art data stream mining has long drawn from ensembles of the Very Fast Decision Tree, a seminal algorithm honored with the 2015 KDD Test-of-Time Award. However, the emergence of large tabular models, i.e., transformers designed for structured numerical data, marks a significant paradigm shift. These models move beyond traditional weight updates, instead employing in-context learning through prompt tuning. By using on-the-fly sketches to summarize unbounded streaming data, one can feed this information into a pre-trained model for efficient processing. This work bridges advancements from both areas, highlighting how transformers' implicit meta-learning abilities, pre-training on drifting natural data, and reliance on context optimization directly address the core challenges of adaptive learning in dynamic environments. Exploring real-time model adaptation, this research demonstrates that TabPFN, coupled with a simple sliding memory strategy, consistently outperforms ensembles of Hoeffding trees, such as Adaptive Random Forest, and Streaming Random Patches, across all non-stationary benchmarks. © 2026 Owner/Author.

Abstract
This paper addresses a multi-compartment tank-truck loading problem for fuel distribution. The proposed problem aims to quantify and assign products to vehicle compartments and to ensure safety throughout the entire distribution using the vehicle Load Distribution Diagram (LDD) to verify vehicle compliance with safety standards and legislation applicable to the transport of dangerous goods. We propose a mixed-integer linear programming model that incorporates axle weight distribution constraints. A new problem generator was developed to test and validate the mathematical model. In the study, three objective functions were considered: minimize operational costs by minimizing the number of compartments allocated to a filling station, maximize profits by maximizing the amount of fuel delivered, and improve safety along the entire route by minimizing the distance between the front of the tank and the load center of gravity. In addition to evaluating these objectives individually, a lexicographic multi-objective approach was implemented to analyse how companies can systematically balance efficiency, profitability, and safety priorities. The computational study demonstrated that LDD constraints are crucial for ensuring the stability and safety of cargo during distribution. Without these constraints, the solutions fail to meet safety standards in 78% of tests. The multi-objective analysis showed limited conflicts among objectives and provided additional managerial insights. Regardless of problem size or objective function, computational times remained consistently low, averaging below 3 seconds.