Abstract
Validation is a central activity when developing formal specifications. Similarly to coding, a possible validation technique is to define upfront test cases or scenarios that a future specification should satisfy or not. Unfortunately, specifying such test cases is burdensome and error prone, which could cause users to skip this validation task. This paper reports the results of an empirical evaluation of using pre-trained large language models (LLMs) to automate the generation of test cases from natural language requirements. In particular, we focus on test cases for structural requirements of simple domain models formalized in the Alloy specification language. Our evaluation focuses on the state-of-the-art GPT-5 model, but results from other closed- and open-source LLMs are also reported. The results show that, in this context, GPT-5 is already quite effective at generating positive (and negative) test cases that are syntactically correct and that satisfy (or not) the given requirement, and that can detect many wrong specifications written by humans. © The Author(s) 2026.

Abstract
Data-driven decision models based on Artificial Intelligence (AI) are increasingly adopted across domains. However, these models are susceptible to bias that can result in unfair or discriminatory outcomes. Recent research has explored causal discovery methods as a promising way to understand and improve fairness in decision-making systems. In this work, we investigate how different conditional independence tests used in constraint-based causal discovery algorithms, specifically the PC algorithm, affect fairness and performance. We perform an empirical evaluation on several datasets, including Portuguese public contracts, COMPAS, and the German Credit dataset. Using seven conditional independence tests, we assess model behavior under fairness (demographic parity, accuracy parity, equalized odds and predictive rate parity) and performance (accuracy, F1-score, AUC) metrics. Our findings reveal that some tests, due to their statistical properties, fail to expose unfairness detectable via causal structures, even when performance metrics appear acceptable. Furthermore, we highlight significant differences in computational efficiency among the tests, with x2-Adf, sp-mi, and sp-x2 being the least efficient. This study underscores the need for careful selection of conditional independence tests in causal discovery to ensure both fairness and reliability in data-driven decision systems.

Abstract
Clinical diagnosis of syndromes benefits strongly from objective facial phenotyping. This study introduces a novel approach to enhance clinical diagnosis through the development and exploration of a low-dimensional metric space referred to as the clinical face phenotypic space (CFPS). As a facial matching tool for clinical genetics, such CFPS can enhance clinical diagnosis. It helps to interpret facial dysmorphisms of a subject by placing them within the space of known dysmorphisms. In this paper, a triplet loss-based autoencoder developed by geometric deep learning (GDL) is trained using multi-task learning, which combines supervised and unsupervised learning approaches. Experiments are designed to illustrate the following properties of CFPSs that can aid clinicians in narrowing down their search space: a CFPS can 1) classify syndromes accurately, 2) generalize to novel syndromes, and 3) preserve the relatedness of genetic diseases, meaning that clusters of phenotypically similar disorders reflect functional relationships between genes. The proposed model consists of three main components: an encoder based on GDL optimizing distances between groups of individuals in the CFPS, a decoder enhancing classification by reconstructing faces, and a singular value decomposition layer maintaining orthogonality and optimal variance distribution across dimensions. This allows for the selection of an optimal number of CFPS dimensions as well as improving the classification capacity of the CFPS, which outperforms the linear metric learning baseline in both syndrome classification and generalization to novel syndromes. We further proved the usefulness of each component of the proposed framework, highlighting their individual impact. From a clinical perspective, the unique combination of these properties in a single CFPS results in a powerful tool that can be incorporated into current clinical practices to assess facial dysmorphism.

Abstract
The energy policy of the European Union stresses the need for sustainable energy consumption, improvements in energy efficiency and lower fossil fuel dependence in a decoupling strategy from unstable democracies. Transportation still represents a sector largely dependent on fossil fuels, which come with several negative impacts. Measuring and assessing the sustainability of the transport sector becomes necessary. This study aims to assess the sustainability performance of the transport sector across 28 European countries over a four-year period, aligned with the policy agenda outlined in strategic documents. The methodological approach involves applying Benefit-of-the-Doubt (BoD) models, comparing aversion that uses transformation methods for anti- isotonic sub-indicators with a variant that directly incorporates these sub-indicators as reverse indicators. In general, the European countries have improved the sustainability performance of their transport sector during the time span analyzed according to the results of both models. For the inefficient units, two improvement strategies are presented based on the profiles identified on the benchmarks from both models, which can be alternative stages to achieve the robust best practices of the benchmarks.

Abstract
Hybrid teaching, which integrates traditional in-person learning based on students' perspectives where online learning offers a flexible approach to education, combines the benefits of technology with face-to-face interactions. Moreover, teaching and learning in a hybrid way met several challenges for both teachers and learners, including technological problems, time management, communication difficulties, and assessment complexities. This systematic review investigates six main research questions: (1) What pedagogical frameworks are used in hybrid teaching and learning? (2) How can we enhance students' engagement in hybrid teaching and learning? (3) What is the impact of technological integration on hybrid learning scenarios, both for students and teachers? (4) How do training and support measures influence the willingness and ability of university teachers to implement hybrid teaching formats? (5) How do formative assessment and feedback methods in hybrid learning environments enable teachers to effectively monitor student progress and provide tailored support? (6) How does the implementation of hybrid learning affect student learning outcomes? This study identifies the following key themes: technological integration, pedagogical innovation, faculty support, student engagement, assessment practices, and learning outcomes. Our contribution of this literature review is related to teaching and learning by showing teachers the most appropriate way to avoid the challenges encountered when teaching in a hybrid way. These include strong technology integration, innovative pedagogical strategies, strong academic development and support, active student engagement, effective assessment practices, and positive learning outcomes.

Abstract
The retention of public transport users is widely acknowledged as a paramount challenge in the path towards the establishment of more sustainable cities and societies. In this setting, in which no contractual relationship with customers exists, an early and accurate prediction of whether a customer will remain with the company or leave, assumes great significance for businesses to develop effective retention strategies. This work focuses on this topic by identifying potential churners based on their past travel behavior. To achieve this, we developed a set of classification models using various machine learning techniques. These models were then employed as base learners within a stacking ensemble. All classifiers were developed with a profit-driven approach, optimizing for expected maximum profit. Finally, we calculated Shapley Additive Explanation values to enhance the interpretability of the proposed classifiers. The performance of the predictive models was evaluated using the data of taxi services recorded in a Portuguese city for 52 months. A broad range of predictors is proposed, including recency and frequency measures of taxi usage as well as others related to customers' satisfaction level. The predictive power of the models was also assessed for specific proportions of higher risk customers. All models have shown the capability to identify churners accurately. This study innovates in evaluating the one-to-one service provider company-customer relationship in the context of taxi industry. Retention actions to promote customers loyalty and enhance retention are also suggested.