Knowledge Graph construction from natural language requires extracting structured triplets from complex, information-dense sentences. In this paper, we investigate if the decomposition of text into atomic propositions (minimal, semantically autonomous units of information) can improve the triplet extraction. We introduce MPropositionneur-V2, a small multilingual model covering six European languages trained by knowledge distillation from Qwen3-32B into a Qwen3-0.6B architecture, and we evaluate its integration into two extraction paradigms: entity-centric (GLiREL) and generative (Qwen3). Experiments on SMiLER, FewRel, DocRED and CaRB show that atomic propositions benefit weaker extractors (GLiREL, CoreNLP, 0.6B models), improving relation recall and, in the multilingual setting, overall accuracy. For stronger LLMs, a fallback combination strategy recovers entity recall losses while preserving the gains in relation extraction. These results show that atomic propositions are an interpretable intermediate data structure that complements extractors without replacing them.

Predicting the 3D structure of RNA is a significant challenge despite ongoing advancements in the field. Although AlphaFold has successfully addressed this problem for proteins, RNA structure prediction raises difficulties due to the fundamental differences between proteins and RNA, which hinder its direct adaptation. The latest release of AlphaFold, AlphaFold3, has broadened its scope to include multiple different molecules such as DNA, ligands and RNA. While the AlphaFold3 article discussed the results for the last CASP-RNA data set, the scope of its performance and the limitations for RNA are unclear. In this article, we provide a comprehensive analysis of the performance of AlphaFold3 in the prediction of 3D structures of RNA. Through an extensive benchmark over five different test sets, we discuss the performance and limitations of AlphaFold3. We also compare its performance with ten existing state-of-the-art ab initio, template-based and deep-learning approaches. Our results are freely available on the EvryRNA platform at https://evryrna.ibisc.univ-evry.fr/evryrna/alphafold3/.

RNA is a molecule that performs critical roles in cellular biology, with its function closely dependent on its three-dimensional conformation. Predicting and evaluating RNA 3D structures remains a significant challenge in the field. Although many metrics and scoring functions have been developed to assess structural quality, each offers a different perspective, and no single method has emerged as a definitive standard. To address this, we previously introduced RNAdvisor, a comprehensive and automated software platform to evaluate 3D RNA structures using a wide range of existing quality metrics and scoring functions. This work presents RNAdvisor2, an extended and improved version of the tool. RNAdvisor 2 introduces a web server designed to enhance accessibility and usability for the broader research community. This release includes new scoring functions and integrates the novel concepts of meta-metrics and meta-scoring functions, which unify diverse evaluation criteria into more robust indicators of RNA structure quality. Additionally, the command-line tool has been improved and optimized for greater stability and extensibility, supporting scalable and maintainable future development. The web server, tool and its source code are freely available on the EvryRNA platform: https://evryrna.ibisc.univ-evry.fr.

In many countries, studies have highlighted the under-representation of women in the media. But beyond quantitative imbalance is the question of the qualitative asymmetry of men and women portrayals. How to help the evaluation of content and salient features specific to male and female discourse? We propose in this study to leverage the knowledge acquired by a classification model trained for gender detection on automatic transcripts, to highlight patterns distinctive of male or female speech. Results show the relevance of coupling explainable AI with classifier confidence to compute consistent attributions.

Knowledge-Aware Visual Question Answering about Entities (KVQAE) is a recent multimodal task aiming to answer visual questions about named entities from a multimodal knowledge base. In this context, we focus more particularly on cross-modal retrieval and propose to inject information about entities in the representations of both texts and images during their building through two pretraining auxiliary tasks, namely entity-level masked language modeling and entity type prediction. We show competitive results over existing approaches on 3 KVQAE standard benchmarks, revealing the benefit of raising entity awareness during cross-modal pretraining, specifically for the KVQAE task.

Large Language Models (LLMs) can be used to produce synthetic documents that mimic real documents when these are not available due to confidentiality or copyright restrictions. Herein, we investigate potential privacy breaches in automatically generated documents. We use synthetic texts generated from a pre-trained model fine-tuned on French clinical cases to evaluate potential privacy breaches according to three directions: (1) similarity between real, training corpus and synthetic corpus (2) strong correlations between clinical features in training and synthetic corpus and (3) Membership Inference Attack (MIA) using a fined tuned model on the synthetic corpus. We identify clinical feature associations that suggest strategies for filtering training corpus that could contribute to privacy preservation. Membership attacks were not conclusive.

Understanding user queries is fundamental in many applications, such as home assistants, booking systems, or recommendations. Accordingly, it is crucial to develop accurate Spoken Language Understanding (SLU) approaches to ensure the reliability of the considered system. Current State-of-the-Art SLU techniques rely on large amounts of training data; however, only limited annotated examples are available for specific tasks or languages. In the meantime, instruction-tuned large language models (LLMs) have shown exceptional performance on unseen tasks in a few-shot setting when provided with adequate prompts. In this work, we propose to explore example selection by leveraging Information retrieval (IR) approaches to build an enhanced prompt that is applied to an SLU task. We evaluate the effectiveness of the proposed method on several SLU benchmarks. Experimental results show that lexical IR methods significantly enhance performance without increasing prompt length.

RNA is a complex macromolecule that plays central roles in the cell. While it is well-known that its structure is directly related to its functions, understanding and predicting RNA structures is challenging. Assessing the real or predictive quality of a structure is also at stake with the complex 3D possible conformations of RNAs. Metrics have been developed to measure model quality while scoring functions aim at assigning quality to guide the discrimination of structures without a known and solved reference. Throughout the years, many metrics and scoring functions have been developed, and no unique assessment is used nowadays. Each developed assessment method has its specificity and might be complementary to understanding structure quality. Therefore, to evaluate RNA 3D structure predictions, it would be important to calculate different metrics and/or scoring functions. For this purpose, we developed RNAdvisor, a comprehensive automated software that integrates and enhances the accessibility of existing metrics and scoring functions. In this paper, we present our RNAdvisor tool, as well as state-of-the-art existing metrics, scoring functions and a set of benchmarks we conducted for evaluating them. Source code is freely available on the EvryRNA platform: https://evryrna.ibisc.univ-evry.fr .

Self-Supervised Learning is vastly used to efficiently represent speech for Spoken Language Understanding, gradually replacing conventional approaches. Meanwhile, textual SSL models are proposed to encode language-agnostic semantics. SAMU-XLSR framework employed this semantic information to enrich multilingual speech representations. A recent study investigated SAMU-XLSR in-domain semantic enrichment by specializing it on downstream transcriptions, leading to state-of-the-art results on a challenging SLU task. This study's interest lies in the loss of multilingual performances and lack of specific-semantics training induced by such specialization in close languages without any SLU implication. We also consider SAMU-XLSR's loss of initial cross-lingual abilities due to a separate SLU fine-tuning. Therefore, this paper proposes a dual task learning approach to improve SAMU-XLSR semantic enrichment while considering distant languages for multilingual and language portability experiments.

The term "speaker diarization" denotes the problem of determining "who speaks when" in a recorded conversation.Among other reasons, it has attracted the attention of the speech research community because of its ability to improve transcription performance, readability and exploitability.Speaker diarization in real-time holds the potential to accelerate and cement the adoption of this technology in our everyday lives.However, although "offline" systems today achieve outstanding performance in pre-recorded conversations, additional problems of "online" real-time diarization, like limited context and low latency, require flexible and efficient solutions enabling both research and production-ready applications.We introduce a Python package called Diart to address real-time speaker diarization in an efficient and flexible way.

This study is part of the debate on the efficiency of large versus small language models for text classification by prompting.We assess the performance of small language models in zero-shot text classification, challenging the prevailing dominance of large models.Across 15 datasets, our investigation benchmarks language models from 77M to 40B parameters using different architectures and scoring functions. Our findings reveal that small models can effectively classify texts, getting on par with or surpassing their larger counterparts.We developed and shared a comprehensive open-source repository that encapsulates our methodologies. This research underscores the notion that bigger isn't always better, suggesting that resource-efficient small models may offer viable solutions for specific data classification challenges.

Predicting the 3D structure of RNA is a significant challenge despite ongoing advancements in the field. Although AlphaFold has successfully addressed this problem for proteins, RNA structure prediction raises difficulties due to fundamental differences between proteins and RNAs, which hinder direct adaptation. The latest release of AlphaFold, AlphaFold 3, has broadened its scope to include multiple different molecules like DNA, ligands and RNA. While the article discusses the results of the last CASP-RNA dataset, the scope of performances and the limitations for RNAs are unclear. In this article, we provide a review of the performance of AlphaFold 3 in the prediction of RNA 3D structures. Through an extensive benchmark over four different test sets, we discuss the performances and limitations of AlphaFold 3. We also compare its performances with ten existing state-of-the-art ab initio, template-based and deep-learning approaches. Our results are freely available on the EvryRNA platform: . ### Competing Interest Statement The authors have declared no competing interest.

MOTIVATION:Predicting the 3D structure of RNA is an ongoing challenge that has yet to be completely addressed despite continuous advancements. RNA 3D structures rely on distances between residues and base interactions but also backbone torsional angles. Knowing the torsional angles for each residue could help reconstruct its global folding, which is what we tackle in this work. This paper presents a novel approach for directly predicting RNA torsional angles from raw sequence data. Our method draws inspiration from the successful application of language models in various domains and adapts them to RNA. RESULTS:We have developed a language-based model, RNA-TorsionBERT, incorporating better sequential interactions for predicting RNA torsional and pseudo-torsional angles from the sequence only. Through extensive benchmarking, we demonstrate that our method improves the prediction of torsional angles compared to state-of-the-art methods. In addition, by using our predictive model, we have inferred a torsion angle-dependent scoring function, called TB-MCQ, that replaces the true reference angles by our model prediction. We show that it accurately evaluates the quality of near-native predicted structures, in terms of RNA backbone torsion angle values. Our work demonstrates promising results, suggesting the potential utility of language models in advancing RNA 3D structure prediction. AVAILABILITY AND IMPLEMENTATION:Source code is freely available on the EvryRNA platform: https://evryrna.ibisc.univ-evry.fr/evryrna/RNA-TorsionBERT.

Intent classification and slot-filling are essential tasks of Spoken Language Understanding (SLU). In most SLUsystems, those tasks are realized by independent modules. For about fifteen years, models achieving both of themjointly and exploiting their mutual enhancement have been proposed. A multilingual module using a joint modelwas envisioned to create a touristic dialogue system for a European project, HumanE-AI-Net. A combination ofmultiple datasets, including the MEDIA dataset, was suggested for training this joint model. The MEDIA SLU datasetis a French dataset distributed since 2005 by ELRA, mainly used by the French research community and free foracademic research since 2020. Unfortunately, it is annotated only in slots but not intents. An enhanced version ofMEDIA annotated with intents has been built to extend its use to more tasks and use cases. This paper presents thesemi-automatic methodology used to obtain this enhanced version. In addition, we present the first results of SLUexperiments on this enhanced dataset using joint models for intent classification and slot-filling.