Abstract
Text readability is vital for effective communication and learning, especially for those with lower information literacy. This research aims to assess Llama 3's ability to grade readability and compare its alignment with established metrics. For that purpose, we create a new dataset of article lead sections from English and Simple English Wikipedia, covering nine categories. The model is prompted to rate the readability of the texts on a grade-level scale, and an in-depth analysis of the results is conducted. While Llama 3 correlates strongly with most metrics, it may underestimate text grade levels.

Abstract
The main challenges limiting the adoption of deep learning-based solutions in medical workflows are the availability of annotated data and the lack of interpretability of such systems. Concept Bottleneck Models (CBMs) tackle the latter by constraining the model output on a set of predefined and human-interpretable concepts. However, the increased interpretability achieved through these concept-based explanations implies a higher annotation burden. Moreover, if a new concept needs to be added, the whole system needs to be retrained. Inspired by the remarkable performance shown by Large Vision-Language Models (LVLMs) in few-shot settings, we propose a simple, yet effective, methodology, CBVLM, which tackles both of the aforementioned challenges. First, for each concept, we prompt the LVLM to answer if the concept is present in the input image. Then, we ask the LVLM to classify the image based on the previous concept predictions. Moreover, in both stages, we incorporate a retrieval module responsible for selecting the best examples for in-context learning. By grounding the final diagnosis on the predicted concepts, we ensure explainability, and by leveraging the few-shot capabilities of LVLMs, we drastically lower the annotation cost. We validate our approach with extensive experiments across four medical datasets and twelve LVLMs (both generic and medical) and show that CBVLM consistently outperforms CBMs and task-specific supervised methods without requiring any training and using just a few annotated examples. More information on our project page: https://cristianopatricio.github.io/CBVLM/.

Abstract
Extreme hypersaline environments harbour a unique biodiversity capable of surviving in such habitats, including halophilic and halotolerant bacteria. Microbial adaptations to these environments comprehend two main strategies: the salt-in that involves a high intracellular concentration of salts (e.g., potassium), and the salt-out that relies on the accumulation of small organic compounds (e.g., glycine betaine and trehalose). These evolutionary haloadaptations, combined with natural population competitiveness, often promotes the production of distinctive antimicrobial compounds, highlighting hypersaline environments as promising rich sources of novel natural products with biotechnological potential. Aiming at enlarging the knowledge on the microbiota of two Portuguese salterns (Aveiro and Olh & atilde;o), microbial isolation was performed using salt and saline sediment samples. A total of 39 microbial isolates were obtained in a saline medium, affiliated with Bacillota, Pseudomonadota, Actinomycetota, and Rhodothermaeota and the archaeal phylum Euryarchaeota. All isolates are generally common in saline habitats, with most (79%) exhibiting a halotolerant profile. Regarding the presence of biosynthetic related genes, 28% of the isolates lacked type I genes for polyketide synthases or non-ribosomal peptide synthetases, 36% contained at least one of these genes, and 36% possessed both. This study provides evidence of the biotechnological potential of the microbiota from two Portuguese salterns.

Abstract
In an era characterized by rapid proliferation of scientific publications and overwhelming volumes of digital content, researchers, students, and faculty members face significant challenges in identifying literature relevant to their academic pursuits. This saturation of information has heightened the need for advanced Recommender Systems within university libraries, tailored specifically for navigating and discovering scientific literature. This paper proposes leveraging insights from librarians' direct interactions with users to adapt existing Recommender Systems, augmented with NLP and LLMs, to better serve the specific needs of academic researchers. It should streamline the research process by delivering precise, relevant, and personalized literature recommendations, centered on a curated database of bibliographic information.

Abstract
In recent years, Positive Energy Districts (PEDs) have been interpreted in many—and often conflicting—ways. We recast PEDs as a vehicle for verifiable climate neutrality and present a declaration-ready assessment that integrates (i) a cumulative, science-based GHG budget per m2 gross floor area (GFA), (ii) full life-cycle accounting, and (iii) time-resolved conversion factors that include everyday motorized individual mobility and quantify flexibility. Two KPIs anchor the framework: the cumulative GHG LCA balance (2025–2075) against a maximum compliant budget of 320 kgCO2e·m-2GFA and the annual primary energy balance used to declare PED status with or without mobility. We follow EN 15978 and apply time-resolved emission factors that decline to zero by 2050. Its applicability is demonstrated on six Austrian districts spanning new builds and renovations, diverse energy systems, densities, and mobility contexts. The baseline scenarios show heterogeneous outcomes—only two out of six meet both the cumulative GHG budget and the positive primary energy balance—but design iterations indicate that all six districts can reach the targets with realistic, ambitious packages (e.g., high energy efficiency and flexibility, local renewables, ecological building materials, BESS/V2G, and mobility electrification). Hourly emission factors and flexibility signals can lower import-weighted emission intensity versus monthly or annual factors by up to 15% and reveal seasonal import–export asymmetries. Built on transparent, auditable rules and open tooling, this framework both diagnoses performance gaps and maps credible pathways to compliance—steering PED design away from project-specific targets toward verifiable climate neutrality. It now serves as the basis for the national labeling/declaration scheme klimaaktiv “Climate-Neutral Positive Energy Districts”. © 2025 by the authors.

Abstract
Deep learning in computational pathology (CPath) has rapidly advanced in recent years. Research has primarily focused on enhancing accuracy and interpretability across various histology image analysis tasks, from tile-level to slide-level foundation models and novel multiple instance learning (MIL) strategies. However, it is equally important for models to provide well-calibrated confidence estimates. Due to factors such as dataset bias, overfitting, and limited training data, existing models tend to be overly confident on test sets. Promising solutions to address this issue include temperature scaling, a post-hoc method that adjusts logits using a single scalar value. However, the role of calibration in CPath is yet to be clarified. In this study, we evaluate temperature scaling and linear temperature scaling for CPath tasks, analyzing their impact on recalibration in both in-domain and out-of-domain distributions. The results show the limitations of current probability calibration techniques and motivate future work.