Abstract
The interaction between humans and technology has always been a key determinant factor of adoption and efficiency. This is true whether the interaction is with hardware, software or data. In the particular case of Information Retrieval (IR), recent developments in Deep Learning and Natural Language Processing (NLP) techniques opened the door to more natural and efficient IR means, no longer based on keywords or similarity metrics but on a distributed representation of meaning. In this paper we propose an agent-based architecture to serve as an interface with industrial systems, in which agents are powered by specific Large Language Models (LLMs). Its main goal is to make the interaction with such systems (e.g. data sources, production systems, machines) natural, allowing users to execute complex tasks with simple prompts. To this end, key aspects considered in the architecture are human-centricity and context-awareness. This paper provides a high-level description of this architecture, and then focuses on the development and evaluation of one of its key agents, responsible for information retrieval. For this purpose, we detail three application scenarios, and evaluate the ability of this agent to select the appropriate data sources to answer a specific prompt. Depending on the scenario and on the underlying model, results show an accuracy of up to 80%, showing that the proposed agent can be used to autonomously select from among several available data sources to answer a specific information need. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

Abstract
This chapter explores the critical role of derivatives in optimizing cost functions and driving the backpropagation algorithm in neural networks, emphasizing their applications in the education field. The study examines the use of derivatives in personalized learning systems, particularly within the Khan Academy platform, and evaluates their impact on scalability, bias, and efficiency. Five research questions guide the analysis, ranging from environmental impact to fairness in AI- driven education. Employing methods like Experimental Performance Evaluation and Comparative Analysis, the study offers both technical insights and ethical considerations. While derivatives enable precise optimization, the chapter highlights how they can unintentionally reinforce biases in training data, raising critical concerns about fairness and representation in educational technologies. © 2025 Elsevier B.V., All rights reserved.

Abstract
This study focuses on exploring new formats and innovative approaches to digital storytelling in museums, offering a critical analysis of existing formats and proposing new perspectives. Initially, current digital storytelling formats are examined, ranging from mobile applications and augmented reality to interactive and multimedia exhibitions. Next, new paradigms and strategies are discussed that aim to expand the possibilities of public engagement and enrich museum experiences. Using a detailed method, careful selections, in-depth analyses and presentation of results are made that highlight both the potential and challenges of these new approaches. The final discussion contextualizes these practices in the current scenario of digital culture and suggests paths for future investigations and developments in the field of digital storytelling in museums. © 2025 Elsevier B.V., All rights reserved.

Abstract
Hydrogen (H-2) is a promising alternative to fossil fuels. However, safety concerns need constant monitoring. Fiber optical sensors have become crucial in this field due to their capability for remote measurements. Traditional plasmonic techniques applied on optical fibers rely on expensive materials, which implies removing the fiber protection, and the optimized bands are outside the infrared spectral range preferred in optical communications. To address these challenges, this work presents an alternative plasmonic structure at the fiber tip of a single-mode fiber. The approach is based on Tamm Plasmon Resonance (TPR), which can be excited at normal incidence with depolarized light. Numerical results indicate that the numerical aperture of the fiber has minimal impact on the TPR band. Experimental results validate the possibility of this approach for H-2 detection, showing a wavelength shift of 8.5nm for 4 vol% H-2 with the TPR band centered around 1565nm. The sensor presents a response time of 29s and a reset time of 27s. These findings open new avenues in the development of plasmonic optical fiber sensors for H-2 sensing, as they enable the possibility of exciting plasmonic modes without removing the fiber's cladding and with simple structures.

Abstract
Context. HD135344AB is a young visual binary system that is best known for the protoplanetary disk around the secondary star. The circumstellar environment of the A0-type primary star, on the other hand, is already depleted. HD135344A is therefore an ideal target for the exploration of recently formed giant planets because it is not obscured by dust. Aims. We searched for and characterized substellar companions to HD135344A down to separations of about 10 au. Methods. We observed HD135344A with VLT/SPHERE in the H23 and K12 bands and obtained YJ and YJH spectroscopy. In addition, we carried out VLTI/GRAVITY observations for the further astrometric and spectroscopic confirmation of a detected companion. Results. We discovered a close-in young giant planet, HD135344Ab, with a mass of about 10 M-J. The multi-epoch astrometry confirms the bound nature based on common parallax and common proper motion. This firmly rules out the scenario of a non-stationary background star. The semi-major axis of the planetary orbit is approximately 15-20 au, and the photometry is consistent with that of a mid L-type object. The inferred atmospheric and bulk parameters further confirm the young and planetary nature of the companion. Conclusions. HD135344Ab is one of the youngest directly imaged planets that has fully formed and orbits on Solar System scales. It is a valuable target for studying the early evolution and atmosphere of a giant planet that could have formed in the vicinity of the snowline.

Abstract
The effectiveness of time series forecasting models can be hampered by conditions in the input space that lead them to underperform. When those are met, negative behaviours, such as higher-than-usual errors or increased uncertainty are shown. Traditionally, stress testing is applied to assess how models respond to adverse, but plausible scenarios, providing insights on how to improve their robustness and reliability. This paper builds upon this technique by contributing with a novel framework called MAST (Meta-learning and data Augmentation for Stress Testing). In particular, MAST is a meta-learning approach that predicts the probability that a given model will perform poorly on a given time series based on a set of statistical features. This way, instead of designing new stress scenarios, this method uses the information provided by instances that led to decreases in forecasting performance. An additional contribution is made, a novel time series data augmentation technique based on oversampling, that improves the information about stress factors in the input space, which elevates the classification capabilities of the method. We conducted experiments using 6 benchmark datasets containing a total of 97.829 time series. The results suggest that MAST is able to identify conditions that lead to large errors effectively.