Abstract
Traditional approaches in the competitive recruitment landscape frequently encounter difficulties in effectively identifying exceptional applicants, resulting in delays, increased expenses, and biases. This study proposes the utilisation of contemporary technologies such as Large Language Models (LLMs) and chatbots to automate the process of resume screening, thereby diminishing prejudices and enhancing communication between recruiters and candidates. Algorithms based on LLM can greatly transform the process of screening by improving both its speed and accuracy. By integrating chatbots, it becomes possible to have personalised interactions with candidates and streamline the process of scheduling interviews. This strategy accelerates the hiring process while maintaining principles of justice and ethics. Its objective is to improve algorithms and procedures to meet changing requirements and enhance the competitive advantage of talent acquisition within organisations.

Abstract
BackgroundTobacco smoke is the main cause of preventable mortality worldwide. Smoking increases the risk of developing many diseases and has been proposed as an aging accelerator. Yet, the molecular mechanisms driving smoking-related health decline and aging acceleration in most tissues remain unexplored.MethodsHere, we use data from the Genotype-Tissue Expression Project (GTEx) to perform a characterization of the effect of cigarette smoking across human tissues. We perform a multi-tissue analysis across 46 human tissues. Our multi-omics characterization includes analysis of gene expression, alternative splicing, DNA methylation, and histological alterations. We further analyze ex-smoker samples to assess the reversibility of these molecular alterations upon smoking cessation.ResultsWe show that smoking impacts tissue architecture and triggers systemic inflammation. We find that in many tissues, the effects of smoking significantly overlap those of aging. Specifically, both age and smoking upregulate inflammatory genes and drive hypomethylation at enhancers (odds ratio (OR) = 2). In addition, we observe widespread smoking-driven hypermethylation at target regions of the Polycomb repressive complex (OR = 2), which is a well-known aging effect. Smoking-induced epigenetic changes overlap causal aging CpGs, suggesting that these methylation changes may directly mediate the aging acceleration observed in smokers. Finally, we find that smoking effects that are shared with aging are more persistent over time.ConclusionOverall, our multi-tissue and multi-omic analysis of the effects of cigarette smoking provides an extensive characterization of the impact of tobacco smoke across tissues and unravels the molecular mechanisms driving smoking-induced tissue homeostasis decline and aging acceleration.

Abstract

Abstract
We introduce a novel multilingual hierarchical corpus annotated for entity framing and role portrayal in news articles. The dataset uses a unique taxonomy inspired by storytelling elements, comprising 22 fine-grained roles, or archetypes, nested within three main categories: protagonist, antagonist, and innocent. Each archetype is carefully defined, capturing nuanced portrayals of entities such as guardian, martyr, and underdog for protagonists; tyrant, deceiver, and bigot for antagonists; and victim, scapegoat, and exploited for innocents. The dataset includes 1,378 recent news articles in five languages (Bulgarian, English, Hindi, European Portuguese, and Russian) focusing on two critical domains of global significance: the Ukraine-Russia War and Climate Change. Over 5,800 entity mentions have been annotated with role labels. This dataset serves as a valuable resource for research into role portrayal and has broader implications for news analysis. We describe the characteristics of the dataset and the annotation process, and we report evaluation results on fine-tuned state-of-the-art multilingual transformers and hierarchical zero-shot learning using LLMs at the level of a document, a paragraph, and a sentence.

Abstract
Obstructive Sleep Apnea (OSA) is a prevalent sleep disorder that significantly affects public health, contributing to cardiovascular and metabolic impairments. Previous studies highlight the heterogeneity of OSA, which is manifested in different phenotypes, complicating personalized treatment strategies. Current phenotyping methods primarily rely on traditional clustering techniques, such as k-means, which may fail to capture complex relationships among features. This study introduces a novel approach based on subject-based SpO(2) weighted correlation networks and modularity analysis to identify clinically relevant subgroups within the OSA population. Using a subset of 2,641 subjects from the Sleep Heart Health Study (SHHS), we extracted 43 SpO(2) features from polysomnography to build correlation networks from them. A bootstrap procedure ensured robustness, while Blondel's modularity algorithm identified subgroups without requiring a predefined number of clusters. Comparison with k-means revealed that the correlation network method identified subgroups with more significantly different sociodemographic, clinical, and anthropometric characteristics (35 variables vs. 28 for k-means). These 35 features effectively revealed hidden SpO2 patterns, suggesting that subject-based correlation networks can identify distinct OSA phenotypes and enhance personalized treatment strategies. This approach improves clinical decisionmaking and patient care. Future research should validate these findings in longitudinal studies and explore integrating multimodal data to refine OSA phenotyping.

Abstract
Detecting anomalies in Waste transportation networks is vital for uncovering illegal or unsafe activities, that can have serious environmental and regulatory consequences. Identifying anomalies in such networks presents a significant challenge due to the limited availability of labeled data and the subtle nature of illicit activities. Moreover, traditional anomaly detection methods relying solely on individual transaction data may overlook deeper, network-level irregularities that arise from complex interactions between entities, especially in the absence of labeled data. This study explores anomaly detection in a waste transport network using unsupervised learning, enhanced by limited supervision and enriched with network structure information. Initially, unsupervised models like Isolation Forest, K-Means, LOF, and Autoencoders were applied using statistical and graph-based features. These models detected outliers without prior labels. Later, information on a few confirmed anomalous users enabled weak supervision, guiding feature selection through statistical tests like Kolmogorov-Smirnov and Anderson-Darling. Results show that models trained on a reduced, graph-focused feature set improved anomaly detection, particularly under extreme class imbalance. Isolation Forest notably ranked known anomalies highly. Ego network visualizations supported these findings, demonstrating the value of integrating structural features and limited labels for identifying subtle, relational anomalies.