Within the current trend of Pretained Language Models (PLM), emerge more and more criticisms about the ethical and ecological impact of such models. In this article, considering these critical remarks, we propose to focus on smaller models, such as compact models like ALBERT, which are more ecologically virtuous than these PLM. However, PLMs enable huge breakthroughs in Natural Language Processing tasks, such as Spoken and Natural Language Understanding, classification, Question–Answering tasks. PLMs also have the advantage of being multilingual, and, as far as we know, a multilingual version of compact ALBERT models does not exist. Considering these facts, we propose the free release of the first version of a multilingual compact ALBERT model, pre-trained using Wikipedia data, which complies with the ethical aspect of such a language model. We also evaluate the model against classical multilingual PLMs in classical NLP tasks. Finally, this paper proposes a rare study on the subword tokenization impact on language performances.

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

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.

La détection d’intention et de concepts sont des tâches essentielles de la compréhension de la parole(SLU). Or il n’existe que peu de données annotées en français permettant d’effectuer ces deux tâches conjointement. Cependant, il existe des ensembles de données annotées en concept, dont le corpus MEDIA. Ce corpus est considéré comme l’un des plus difficiles. Néanmoins, il ne comporte que des annotations en concepts et pas en intentions. Dans cet article, nous proposons une version étendue de MEDIA annotée en intentions pour étendre son utilisation. Cet article présente une méthode semi-automatique pour obtenir cette version étendue. De plus, nous présentons les premiers résultats des expériences menées sur cet ensemble de données en utilisant des modèles joints pour la classification des intentions et la détection de concepts.

Ce travail s’inscrit dans le débat sur l’efficacité des grands modèles de langue par rapport aux petits pour la classification de texte par amorçage (prompting). Nous évaluons ici le potentiel des petits modèles de langue dans la classification de texte sans exemples, remettant en question la prédominance des grands modèles. À travers un ensemble diversifié de jeux de données, notre étude compare les petits et les grands modèles utilisant différentes architectures et données de pré-entraînement. Nos conclusions révèlent que les petits modèles peuvent générer efficacement des étiquettes et, dans certains contextes, rivaliser ou surpasser les performances de leurs homologues plus grands. Ce travail souligne l’idée que le modèle le plus grand n’est pas toujours le meilleur, suggérant que les petits modèles économes en ressources peuvent offrir des solutions viables pour des défis spécifiques de classification de données

Dans de nombreux pays, des études ont souligné la sous-représentation des femmes dans les médias.Mais au-delà du déséquilibre quantitatif se pose la question de l’asymétrie qualitative des représentations des hommes et des femmes.Comment automatiser l’évaluation des contenus et des traits saillants spécifiques aux discours masculins et féminins ?Nous proposons dans cette étude d’exploiter les connaissances acquises par un modèle de classification entraîné à la détection du genre sur des transcriptions automatiques, afin de mettre en évidence des motifs distinctifs du discours masculin ou féminin.Notre approche est basée sur l’utilisation de méthodes développées pour l’intelligence artificielle explicable (IAX), afin de calculer des scores d’attribution au niveau des unités.

Le projet ANR Gender Equality Monitor (GEM) est coordonné par l’Institut National de l’Audiovisuel(INA) et vise à étudier la place des femmes dans les médias (radio et télévision). Dans cette soumission,nous présentons le travail réalisé au LISN : (i) étude diachronique des caractéristiques acoustiquesde la voix en fonction du genre et de l’âge, (ii) comparaison acoustique de la voix des femmeset hommes politiques montrant une incohérence entre performance vocale et commentaires sur lavoix, (iii) réalisation d’un système automatique d’estimation de la féminité perçue à partir descaractéristiques vocales, (iv) comparaison de systèmes de segmentation thématique de transcriptionsautomatiques de données audiovisuelles, (v) mesure des biais sociétaux dans les modèles de languedans un contexte multilingue et multi-culturel, et (vi) premiers essais d’identification de la publicitéen fonction du genre du locuteur.

Knowledge transfer between neural language models is a widely used technique that has proven to improve performance in a multitude of natural language tasks, in particular with the recent rise of large pre-trained language models like BERT. Similarly, high cross-lingual transfer has been shown to occur in multilingual language models. Hence, it is of great importance to better understand this phenomenon as well as its limits. While most studies about cross-lingual transfer focus on training on independent and identically distributed (i.e. i.i.d.) samples, in this paper we study cross-lingual transfer in a continual learning setting on two sequence labeling tasks: slot-filling and named entity recognition. We investigate this by training multilingual BERT on sequences of 9 languages, one language at a time, on the MultiATIS++ and MultiCoNER corpora. Our first findings are that forward transfer between languages is retained although forgetting is present. Additional experiments show that lost performance can be recovered with as little as a single training epoch even if forgetting was high, which can be explained by a progressive shift of model parameters towards a better multilingual initialization. We also find that commonly used metrics might be insufficient to assess continual learning performance.

With the emergence of neural end-to-end approaches for spoken language understanding (SLU), a growing number of studies have been presented during these last three years on this topic. The major part of these works addresses the spoken language understanding domain through a simple task like speech intent detection. In this context, new benchmark datasets have also been produced and shared with the community related to this task. In this paper, we focus on the French MEDIA SLU dataset, distributed since 2005 and used as a benchmark dataset for a large number of research works. This dataset has been shown as being the most challenging one among those accessible to the research community. Distributed by ELRA, this corpus is free for academic research since 2019. Unfortunately, the MEDIA dataset is not really used beyond the French research community. To facilitate its use, a complete recipe, including data preparation, training and evaluation scripts, has been built and integrated to SpeechBrain, an already popular open-source and all-in-one conversational AI toolkit based on PyTorch. This recipe is presented in this paper. In addition, based on the feedback of some researchers who have worked on this dataset for several years, some corrections have been brought to the initial manual annotation: the new version of the data will also be integrated into the ELRA catalogue, as the original one. More, a significant amount of data collected during the construction of the MEDIA corpus in the 2000s was never used until now: we present the first results reached on this subset — also included in the MEDIA SpeechBrain recipe — , that will be used for now as the MEDIA test2. Last, we discuss evaluation issues.

Pretrained models through self-supervised learning have been recently introduced for both acoustic and language modeling. Applied to spoken language understanding tasks, these models have shown their great potential by improving the state-of-the-art performances on challenging benchmark datasets. In this paper, we present an error analysis reached by the use of such models on the French MEDIA benchmark dataset, known as being one of the most challenging benchmarks for the slot filling task among all the benchmarks accessible to the entire research community. One year ago, the state-of-art system reached a Concept Error Rate (CER) of 13.6% through the use of a end-to-end neural architecture. Some months later, a cascade approach based on the sequential use of a fine-tuned wav2vec2.0 model and a fine-tuned BERT model reaches a CER of 11.2%. This significant improvement raises questions about the type of errors that remain difficult to treat, but also about those that have been corrected using these models pre-trained through self-supervision learning on a large amount of data. This study brings some answers in order to better understand the limits of such models and open new perspectives to continue improving the performance.

Modern Natural Language Processing (NLP) makes intensive use of deep learning methods because of the accuracy they offer for a variety of applications. Due to the significant environmental impact of deep learning, cost-benefit analysis including carbon footprint as well as accuracy measures has been suggested to better document the use of NLP methods for research or deployment. In this paper, we review the tools that are available to measure energy use and CO2 emissions of NLP methods. We describe the scope of the measures provided and compare the use of six tools (carbon tracker, experiment impact tracker, green algorithms, ML CO2 impact, energy usage and cumulator) on named entity recognition experiments performed on different computational set-ups (local server vs. computing facility). Based on these findings, we propose actionable recommendations to accurately measure the environmental impact of NLP experiments.

In this paper, we present a study on a French Spoken Language Understanding (SLU) task: the MEDIA task. Many works and studies have been proposed for many tasks, but most of them are focused on English language and tasks. The exploration of a richer language like French within the framework of a SLU task implies to recent approaches to handle this difficulty. Since the MEDIA task seems to be one of the most difficult, according several previous studies, we propose to explore Neural Networks approaches focusing of three aspects: firstly, the Neural Network inputs and more specifically the word embeddings; secondly, we compared French version of BERT against the best setup through different ways; Finally, the comparison against State-of-the-Art approaches. Results show that the word embeddings trained on a small corpus need to be updated during SLU model training. Furthermore, the French BERT fine-tuned approaches outperform the classical Neural Network Architectures and achieves state of the art results. However, the contextual embeddings extracted from one of the French BERT approaches achieve comparable results in comparison to word embedding, when integrated into the proposed neural architecture.

This paper describes the participation of LIMSI_UPV team in SemEval-2020 Task 9: Sentiment Analysis for Code-Mixed Social Media Text. The proposed approach competed in SentiMix HindiEnglish subtask, that addresses the problem of predicting the sentiment of a given Hindi-English code-mixed tweet. We propose Recurrent Convolutional Neural Network that combines both the recurrent neural network and the convolutional network to better capture the semantics of the text, for code-mixed sentiment analysis. The proposed system obtained 0.69 (best run) in terms of F1 score on the given test data and achieved the 9th place (Codalab username: somban) in the SentiMix Hindi-English subtask.

The task of automatic misogyny identification and categorization has not received as much attention as other natural language tasks have, even though it is crucial for identifying hate speech in social Internet interactions. In this work, we address this sentence classification task from a representation learning perspective, using both a bidirectional LSTM and BERT optimized with the following metric learning loss functions: contrastive loss, triplet loss, center loss, congenerous cosine loss and additive angular margin loss. We set new state-of-the-art for the task with our fine-tuned BERT, whose sentence embeddings can be compared with a simple cosine distance, and we release all our code as open source for easy reproducibility. Moreover, we find that almost every loss function performs equally well in this setting, matching the regular cross entropy loss.

Récemment, l’utilisation des représentations continues de mots a connu beaucoup de succès dans plusieurs tâches de traitement du langage naturel. Dans cet article, nous proposons d’étudier leur utilisation dans une architecture neuronale pour la tâche de détection des erreurs au sein de transcriptions automatiques de la parole. Nous avons également expérimenté et évalué l’utilisation de paramètres prosodiques en suppléments des paramètres classiques (lexicaux, syntaxiques, . . .). La principale contribution de cet article porte sur la combinaison de différentes représentations continues de mots : plusieurs approches de combinaison sont proposées et évaluées afin de tirer profit de leurs complémentarités. Les expériences sont effectuées sur des transcriptions automatiques du corpus ETAPE générées par le système de reconnaissance automatique du LIUM. Les résultats obtenus sont meilleurs que ceux d’un système état de l’art basé sur les champs aléatoires conditionnels. Pour terminer, nous montrons que la mesure de confiance produite est particulièrement bien calibrée selon une évaluation en terme d’Entropie Croisée Normalisée (NCE).

Word embeddings have been successfully used in several natural language processing tasks (NLP) and speech processing. Different approaches have been introduced to calculate word embeddings through neural networks. In the literature, many studies focused on word embedding evaluation, but for our knowledge, there are still some gaps. This paper presents a study focusing on a rigorous comparison of the performances of different kinds of word embeddings. These performances are evaluated on different NLP and linguistic tasks, while all the word embeddings are estimated on the same training data using the same vocabulary, the same number of dimensions, and other similar characteristics. The evaluation results reported in this paper match those in the literature, since they point out that the improvements achieved by a word embedding in one task are not consistently observed across all tasks. For that reason, this paper investigates and evaluates approaches to combine word embeddings in order to take advantage of their complementarity, and to look for the effective word embeddings that can achieve good performances on all tasks. As a conclusion, this paper provides new perceptions of intrinsic qualities of the famous word embedding families, which can be different from the ones provided by works previously published in the scientific literature.