Abstract

Abstract
To fulfil their mission, museums strive to ensure a balance between the inclusive fruition of collections and preservation. In this sense, as part of their non-formal educational functions, some try to develop multimedia resources to explore content, engage, and promote knowledge. This paper is based on a gamified mobile application developed for a museum in Portugal, targeting young people between 8 and 12 years old, focusing on the evaluation phase of the experience’s impact. To this end, the Method for Assessing eXperience (MAX) was adopted. MAX is presented, and introduced changes are shared, in order to simplify its application. Using MAX’s simplified version, four sessions were established with a target group of 44 young museum visitors. Results show an overall satisfaction towards the gamified mobile application. It is possible to understand that 68% of the 44 young people considered the application very useful and 57% expressed the desire to use it again in the near future. Regarding emotions during the experience, a mere 2% reported confusion, while an overwhelming 98% showcased positive reactions. This study concludes that the application may have contributed positively to raising awareness for the preservation of cultural heritage and this assessment has enabled the overall improvement of the prototype. It is believed that the provided fun involvement might help to develop further knowledge regarding museum practices, supporting and enhancing info-communicational flows. © 2025 Elsevier B.V., All rights reserved.

Abstract
This article analyzes the impact of Communities of Practice (CoP) in the context of professional growth and pedagogical development in higher education in the wake of Generative Artificial Intelligence (GAI) technology. Based on the INOV-NORTE initiative and concentrating on CAP Generative Artificial Intelligence and Higher Education: First Steps, the research addresses the problem of how faculty can be supported through peer organization to critically and creatively apply GAI into teaching through structured, unlockable teaching frameworks. Using qualitative case study methodology, the article captures participants ' engagements with AI through ethical, technological, and pedagogical lenses, analysis, and policy curation, collaborative debates, practical workshops, and teamwork that occurred both synchronously and asynchronously. Results underscore the CoP's contribution to fostering interdisciplinary engagement, including informed ethical thinking, innovation, and proactive curricular change, along with other barriers to participating in digitally enabled governance and institutional responsiveness. The research validates the integration of CoP as professional learning frameworks that invite multiple perspectives and scales responsive to various contexts, illustrating the urgent need for institutions to strategically adapt their policies to remain relevant in the rapidly evolving educational landscape shaped by AI technology.

Abstract
The European Project Semester (EPS), offered by the Instituto Superior de Engenharia do Porto (ISEP), is a capstone programme designed for undergraduate students in engineering, product design, and business. EPS@ISEP fosters project-based learning, promotes multicultural and interdisciplinary teamwork, and ethics- and sustainability-driven design. This study applies Natural Language Processing techniques, specifically text mining, to analyse project papers produced by EPS@ISEP teams. The proposed method aims to identify evidence of ethical concerns within EPS@ISEP projects. An innovative keyword mapping approach is introduced that first defines and refines a list of ethics-related keywords through prompt engineering. This enriched list of keywords is then used to systematically map the content of project papers. The findings indicate that the EPS@ISEP robotics project papers analysed demonstrate awareness of ethical considerations and actively incorporate them into design processes. The method presented is adaptable to various application areas, such as monitoring compliance with responsible innovation or sustainability policies.

Abstract
This paper introduces PRISMA, a hybrid multi-agent Deep Reinforcement Learning (DRL) framework for solving the Flexible Job-shop Scheduling Problem (FJSP). It uses Constraint Programming (CP) solutions to pretrain decentralized policies and to guide exploration during training. Although DRL can generate fast solutions for large combinatorial problems, it often fails to match the quality of optimization methods, motivating the integration with hybrid frameworks. The growing interest in embedding domain knowledge into learning algorithms has produced several hybrid formulations, yet their potential remains underexplored, particularly in multi-agent settings. PRISMA combines supervised and reinforcement learning within a multi-agent framework, where CP solutions are used to (i) learn expert decisions through imitation learning, and (ii) train an auxiliary network that guides DRL training via reward shaping. A shared graph network is adopted for transferring system-level knowledge into machine-level observations, enabling fast and consistent inference from enriched local embed-dings. To the best of our knowledge, PRISMA introduces the first expert-derived guidance mechanism for the FJSP and is among the earliest to apply imitation learning within a multi-agent formulation. By combining both modules, it strengthens the bridge between optimization and learning-based methods, where such dual integrations remain scarce. Experimental results show faster convergence and higher solution quality than state-ofthe-art DRL models. PRISMA achieves an average optimality gap of 6.74%, corresponding to a 50% relative improvement over the single-agent baseline, while reducing inference time. These findings reinforce the value of merging optimization accuracy with the flexibility of multi-agent DRL for efficient scheduling.

Abstract
Real-world datasets used in data-driven modelling are often affected by inconsistencies arising from discrepancies between recorded observations and actual system behaviour. Conventional approaches to data filtering and quality assessment rely primarily on statistical criteria and treat consistency as an intrinsic property of the dataset. However, such approaches may fail to identify observations that are statistically plausible yet incompatible with the structural assumptions underlying the model.This work introduces a model-dependent perspective on data consistency, in which the validity of observations is defined relative to the model used to interpret the data rather than to distributional properties alone. Within this framework, residuals are interpreted not merely as noise, but as indicators of incompatibility between observed data and model-defined behaviour.Importantly, inconsistencies may arise not only from anomalies in the target variable, but also from incorrect, incomplete, or misaligned representations of explanatory variables, even when observed outputs remain statistically valid. By formalising data consistency as a model-dependent property, this work challenges the conventional separation between data preprocessing and modelling, and reframes data filtering as part of the interpretation of model-data relations.The proposed framework provides a conceptual basis for integrating data consistency into data-driven modelling processes, with implications for data interpretation, representation, and validation in systems operating under imperfect real-world data conditions.