Recent surge in the accessibility of large language models (LLMs) to the general population can lead to untrackable use of such models for medical-related recommendations. Language generation via LLMs models has two key problems: firstly, they are prone to hallucination and therefore, for any medical purpose they require scientific and factual grounding; secondly, LLMs pose tremendous challenge to computational resources due to their gigantic model size. In this work, we introduce pRAGe, a Pipeline for Retrieval Augmented Generation and Evaluation of medical paraphrases generation using Small Language Models (SLM). We study the effectiveness of SLMs and the impact of external knowledge base for medical paraphrase generation in French.

Cet article présente une étude sur l’utilisation de modèles de prédiction de liens pour l’enrichissement de graphes lexico-sémantiques du français. Celle-ci porte sur deux graphes, RezoJDM16k et RL-fr et sept modèles de prédiction de liens. Nous avons étudié les prédictions du modèle le plus performant, afin d’extraire de potentiels nouveaux triplets en utilisant un score de confiance que nous avons évalué avec des annotations manuelles. Nos résultats mettent en évidence des avantages différentspour le graphe dense RezoJDM16k par rapport à RL-fr, plus clairsemé. Si l’ajout de nouveaux triplets à RezoJDM16k offre des avantages limités, RL-fr peut bénéficier substantiellement de notre approche.

Cette étude s’intéresse à la prédiction de la complexité lexicale. Nous explorons des méthodesd’apprentissage profond afin d’évaluer la complexité d’un mot en se basant sur son contexte. Plusspécifiquement, nous examinons comment utiliser des modèles de langue pré-entraînés pour encoderle mot cible et son contexte, en les combinant avec des caractéristiques supplémentaires basées sur lafréquence. Notre approche obtient de meilleurs résultats que les meilleurs systèmes de SemEval-2021(Shardlow et al., 2021). Enfin, nous menons une étude comparative avec ChatGPT afin d’évaluer sonpotentiel pour prédire la complexité lexicale en comparaison avec un modèle dédié à cette tâche.

The success of pretrained language models (PLMs) across a spate of use-cases has led to significant investment from the NLP community towards building domain-specific foundational models. On the other hand, in mission critical settings such as biomedical applications, other aspects also factor in—chief of which is a model’s ability to produce reasonable estimates of its own uncertainty. In the present study, we discuss these two desiderata through the lens of how they shape the entropy of a model’s output probability distribution. We find that domain specificity and uncertainty awareness can often be successfully combined, but the exact task at hand weighs in much more strongly.

This paper presents a resource-centric study of link prediction approaches over French lexical-semantic graphs. Our study incorporates two graphs, RezoJDM16k and RL-fr, and we evaluated seven link prediction models, with CompGCN-ConvE emerging as the best performer. We also conducted a qualitative analysis of the predictions using manual annotations. Based on this, we found that predictions with higher confidence scores were more valid for inclusion. Our findings highlight different benefits for the dense graph compared to the sparser graph RL-fr. While the addition of new triples to RezoJDM16k offers limited advantages, RL-fr can benefit substantially from our approach.

There are several works in natural language processing for identifying lexical complexity. This can be for various reasons, either for simplification, the selection of more suitable content, or for other specific tasks. Words can have multiple definitions and degrees of complexity depending on the context in which they appear. One solution being investigated is lexical complexity prediction, where computational methods are used to evaluate the difficulty of vocabulary for language learners and offer personalized assistance. In this work, we explore deep learning methods to assess the complexity of a word based on its context. Specifically, we investigate how to use pre-trained language models to encode both the sentence and the target word, and then fine-tune them by combining them with additional frequency-based features. Our approach achieved superior results compared to the best systems in SemEval-2021 (Shardlow et al., 2021), as demonstrated by an R2 score of 0.65. Finally, we carry out a comparative study with ChatGPT to assess its potential for predicting lexical complexity, to see whether prompt engineering can be an alternative to this task, we will discuss the advantages and limitations of ChatGPT.

Au début du XXIe siècle, le français faisait encore partie des langues peu dotées. Grâce aux efforts de la communauté française du traitement automatique des langues (TAL), de nombreuses ressources librement disponibles ont été produites, dont des lexiques du français. À travers cet article, nous nous intéressons à leur devenir dans la communauté par le prisme des actes de la conférence TALN sur une période de 20 ans.

The number of scientific articles is increasing tremendously across all domains to such an extent that it has become hard for researchers to remain up-to-date. Evidently, scientific language understanding systems and Information Extraction (IE) systems, with the advancement of Natural Language Processing (NLP) techniques, are benefiting the needs of users. Although the majority of the practices for building such systems are data-driven, advocating the idea of “The more, the better”. In this work, we revisit the paradigm - questioning what type of data : text (title, abstract) or citations, can have more impact on the performance of scientific language understanding systems.

How does the word analogy task fit in the modern NLP landscape? Given the rarity of comparable multilingual benchmarks and the lack of a consensual evaluation protocol for contextual models, this remains an open question. In this paper, we introduce MATS: a multilingual analogy dataset, covering forty analogical relations in six languages, and evaluate human as well as static and contextual embedding performances on the task. We find that not all analogical relations are equally straightforward for humans, static models remain competitive with contextual embeddings, and optimal settings vary across languages and analogical relations. Several key challenges remain, including creating benchmarks that align with human reasoning and understanding what drives differences across methodologies.

Word embeddings have advanced the state of the art in NLP across numerous tasks. Understanding the contents of dense neural representations is of utmost interest to the computational semantics community. We propose to focus on relating these opaque word vectors with human-readable definitions, as found in dictionaries This problem naturally divides into two subtasks: converting definitions into embeddings, and converting embeddings into definitions. This task was conducted in a multilingual setting, using comparable sets of embeddings trained homogeneously.

This paper focuses on the task of word sense disambiguation (WSD) on lexicographic examples relying on the French Lexical Network (fr-LN). For this purpose, we exploit the lexical and relational properties of the network, that we integrated in a feedforward neural WSD model on top of pretrained French BERT embeddings. We provide a comparative study with various models and further show the impact of our approach regarding polysemic units.

Pretrained embeddings based on the Transformer architecture have taken the NLP community by storm. We show that they can mathematically be reframed as a sum of vector factors and showcase how to use this reframing to study the impact of each component. We provide evidence that multi-head attentions and feed-forwards are not equally useful in all downstream applications, as well as a quantitative overview of the effects of finetuning on the overall embedding space. This approach allows us to draw connections to a wide range of previous studies, from vector space anisotropy to attention weights.

Biomedical NER is an active research area today. Despite the availability of state-of-the-art models for standard NER tasks, their performance degrades on biomedical data due to OOV entities and the challenges encountered in specialized domains. We use Flair-NER framework to investigate the effectiveness of various contextual and static embeddings for NER on Spanish tweets, in particular, to capture complex disease mentions.

Pour les langues historiques non stabilisées comme le français médiéval, la lemmatisation automatique présente toujours des défis, car cette langue connaît une forte variation graphique. Dans cet article, nous dressons un état des lieux de la lemmatisation automatique pour cette langue en comparant les performances de quatre lemmatiseurs existants sur un même jeu de données. L’objectif est d’évaluer où se situent les nouvelles techniques de l’apprentissage automatique par rapport aux techniques plus traditionnelles s’appuyant sur des systèmes de règles et lexiques, en particulier pour la prédiction des mots inconnus.

We present here Rigor Mortis, a gamified crowdsourcing platform designed to evaluate the intuition of the speakers, then train them to annotate multi-word expressions (MWEs) in French corpora. We previously showed that the speakers’ intuition is reasonably good (65% in recall on non-fixed MWE). We detail here the annotation results, after a training phase using some of the tests developed in the PARSEME-FR project.

La génération de définitions est une tâche récente qui vise à produire des définitions lexicographiques à partir de plongements lexicaux. Nous remarquons deux lacunes : (i) l’état de l’art actuel ne s’est penché que sur l’anglais et le chinois, et (ii) l’utilisation escomptée en tant que méthode d’évaluation des plongements lexicaux doit encore être vérifiée. Pour y remédier, nous proposons un jeu de données pour la génération de définitions en français, ainsi qu’une évaluation des performances d’un modèle de génération de définitions simple selon les plongements lexicaux fournis en entrée.

Nous présentons le démonstrateur en-ligne du projet ANR PARSEME-FR dédié aux expressions polylexicales. Il inclut différents outils d’identification de telles expressions et un outil d’exploration des ressources linguistiques de ce projet.

This article focuses on the lemmatization of multiword expressions (MWEs). We propose a deep encoder-decoder architecture generating for every MWE word its corresponding part in the lemma, based on the internal context of the MWE. The encoder relies on recurrent networks based on (1) the character sequence of the individual words to capture their morphological properties, and (2) the word sequence of the MWE to capture lexical and syntactic properties. The decoder in charge of generating the corresponding part of the lemma for each word of the MWE is based on a classical character-level attention-based recurrent model. Our model is evaluated for Italian, French, Polish and Portuguese and shows good performances except for Polish.

In this paper, we compare the use of linear versus neural classifiers in a greedy transition system for MWE identification. Both our linear and neural models achieve a new state-of-the-art on the PARSEME 1.1 shared task data sets, comprising 20 languages. Surprisingly, our best model is a simple feed-forward network with one hidden layer, although more sophisticated (recurrent) architectures were tested. The feedback from this study is that tuning a SVM is rather straightforward, whereas tuning our neural system revealed more challenging. Given the number of languages and the variety of linguistic phenomena to handle for the MWE identification task, we have designed an accurate tuning procedure, and we show that hyperparameters are better selected by using a majority-vote within random search configurations rather than a simple best configuration selection. Although the performance is rather good (better than both the best shared task system and the average of the best per-language results), further work is needed to improve the generalization power, especially on unseen MWEs.

Defining words in a textual context is a useful task both for practical purposes and for gaining insight into distributed word representations. Building on the distributional hypothesis, we argue here that the most natural formalization of definition modeling is to treat it as a sequence-to-sequence task, rather than a word-to-sequence task: given an input sequence with a highlighted word, generate a contextually appropriate definition for it. We implement this approach in a Transformer-based sequence-to-sequence model. Our proposal allows to train contextualization and definition generation in an end-to-end fashion, which is a conceptual improvement over earlier works. We achieve state-of-the-art results both in contextual and non-contextual definition modeling.

This article presents the results we obtained in crowdsourcing French speakers’ intuition concerning multi-work expressions (MWEs). We developed a slightly gamified crowdsourcing platform, part of which is designed to test users’ ability to identify MWEs with no prior training. The participants perform relatively well at the task, with a recall reaching 65% for MWEs that do not behave as function words.

Multiword expressions (MWEs) are a class of linguistic forms spanning conventional word boundaries that are both idiosyncratic and pervasive across different languages. The structure of linguistic processing that depends on the clear distinction between words and phrases has to be re-thought to accommodate MWEs. The issue of MWE handling is crucial for NLP applications, where it raises a number of challenges. The emergence of solutions in the absence of guiding principles motivates this survey, whose aim is not only to provide a focused review of MWE processing, but also to clarify the nature of interactions between MWE processing and downstream applications. We propose a conceptual framework within which challenges and research contributions can be positioned. It offers a shared understanding of what is meant by “MWE processing,” distinguishing the subtasks of MWE discovery and identification. It also elucidates the interactions between MWE processing and two use cases: Parsing and machine translation. Many of the approaches in the literature can be differentiated according to how MWE processing is timed with respect to underlying use cases. We discuss how such orchestration choices affect the scope of MWE-aware systems. For each of the two MWE processing subtasks and for each of the two use cases, we conclude on open issues and research perspectives.

We describe the ATILF-LLF system built for the MWE 2017 Shared Task on automatic identification of verbal multiword expressions. We participated in the closed track only, for all the 18 available languages. Our system is a robust greedy transition-based system, in which MWE are identified through a MERGE transition. The system was meant to accommodate the variety of linguistic resources provided for each language, in terms of accompanying morphological and syntactic information. Using per-MWE Fscore, the system was ranked first for all but two languages (Hungarian and Romanian).

This article evaluates the extension of a dependency parser that performs joint syntactic analysis and multiword expression identification. We show that, given sufficient training data, the parser benefits from explicit multiword information and improves overall labeled accuracy score in eight of the ten evaluation cases.

Nous décrivons la partie française des données produites dans le cadre de la campagne multilingue PARSEME sur l’identification d’expressions polylexicales verbales (Savary et al., 2017). Les expressions couvertes pour le français sont les expressions verbales idiomatiques, les verbes intrinsèquement pronominaux et une généralisation des constructions à verbe support. Ces phénomènes ont été annotés sur le corpus French-UD (Nivre et al., 2016) et le corpus Sequoia (Candito & Seddah, 2012), soit un corpus de 22 645 phrases, pour un total de 4 962 expressions annotées. On obtient un ratio d’une expression annotée tous les 100 tokens environ, avec un fort taux d’expressions discontinues (40%).

Verbenet is a French lexicon developed by “translation” of its English counterpart — VerbNet (Kipper-Schuler, 2005)—and treatment of the specificities of French syntax (Pradet et al., 2014; Danlos et al., 2016). One difficulty encountered in its development springs from the fact that the list of (potentially numerous) frames has no internal organization. This paper proposes a type system for frames that shows whether two frames are variants of a given alternation. Frame typing facilitates coherence checking of the resource in a “virtuous circle”. We present the principles underlying a program we developed and used to automatically type frames in VerbeNet. We also show that our system is portable to other languages.

L’article présente des résultats d’expériences d’apprentissage automatique pour l’étiquetage morpho-syntaxique et l’analyse syntaxique en dépendance de l’ancien français. Ces expériences ont pour objectif de servir une exploration de corpus pour laquelle le corpus arboré SRCMF sert de données de référence. La nature peu standardisée de la langue qui y est utilisée implique des données d’entraînement hétérogènes et quantitativement limitées. Nous explorons donc diverses stratégies, fondées sur différents critères (variabilité du lexique, forme Vers/Prose des textes, dates des textes), pour constituer des corpus d’entrainement menant aux meilleurs résultats possibles.

In this article, we describe a new sense-tagged corpus for Word Sense Disambiguation. The corpus is constituted of instances of 20 French polysemous verbs. Each verb instance is annotated with three sense labels: (1) the actual translation of the verb in the english version of this instance in a parallel corpus, (2) an entry of the verb in a computational dictionary of French (the Lexicon-Grammar tables) and (3) a fine-grained sense label resulting from the concatenation of the translation and the Lexicon-Grammar entry.

This paper evaluates the impact of external lexical resources into a CRF-based joint Multiword Segmenter and Part-of-Speech Tagger. We especially show different ways of integrating lexicon-based features in the tagging model. We display an absolute gain of 0.5% in terms of f-measure. Moreover, we show that the integration of lexicon-based features significantly compensates the use of a small training corpus.

We present an extension of the adverbial entries of the French morphological lexicon DELA (Dictionnaires Electroniques du LADL / LADL electronic dictionaries). Adverbs were extracted from LGLex, a NLP-oriented syntactic resource for French, which in its turn contains all adverbs extracted from the Lexicon-Grammar tables of both simple adverbs ending in -ment (i.e., '-ly') and compound adverbs. This work exploits fine-grained linguistic information provided in existing resources. The resulting resource is reviewed in order to delete duplicates and is freely available under the LGPL-LR license.

Dans cet article, nous synthétisons les résultats de plusieurs séries d’expériences réalisées à l’aide de CRF (Conditional Random Fields ou “champs markoviens conditionnels”) linéaires pour apprendre à annoter des textes français à partir d’exemples, en exploitant diverses ressources linguistiques externes. Ces expériences ont porté sur l’étiquetage morphosyntaxique intégrant l’identification des unités polylexicales. Nous montrons que le modèle des CRF est capable d’intégrer des ressources lexicales riches en unités multi-mots de différentes manières et permet d’atteindre ainsi le meilleur taux de correction d’étiquetage actuel pour le français.

This paper describes the process and the resources used to automatically annotate a French corpus of spontaneous speech transcriptions in super-chunks. Super-chunks are enhanced chunks that can contain lexical multiword units. This partial parsing is based on a preprocessing stage of the spoken data that consists in reformatting and tagging utterances that break the syntactic structure of the text, such as disfluencies. Spoken specificities were formalized thanks to a systematic linguistic study of a 40-hour-long speech transcription corpus. The chunker uses large-coverage and fine-grained language resources for general written language that have been augmented with resources specific to spoken French. It consists in iteratively applying finite-state lexical and syntactic resources and outputing a finite automaton representing all possible chunk analyses. The best path is then selected thanks to a hybrid disambiguation stage. We show that our system reaches scores that are comparable with state-of-the-art results in the field.

In this article, we present an experiment of linguistic parameter tuning in the representation of the semantic space of polysemous words. We evaluate quantitatively the influence of some basic linguistic knowledge (lemmas, multi-word expressions, grammatical tags and syntactic relations) on the performances of a similarity-based Word-Sense disambiguation method. The question we try to answer, by this experiment, is which kinds of linguistic knowledge are most useful for the semantic disambiguation of polysemous words, in a multilingual framework. The experiment is about 20 French polysemous words (16 nouns and 4 verbs) and we make use of the French-English part of the sentence-aligned EuroParl Corpus for training and testing. Our results show a strong correlation between the system accuracy and the degree of precision of the linguistic features used, particularly the syntactic dependency relations. Furthermore, the lemma-based approach absolutely outperforms the word form-based approach. The best accuracy achieved by our system amounts to 90%.

Depuis l’analyseur développé par Harris à la fin des années 50, les unités polylexicales ont peu à peu été intégrées aux analyseurs syntaxiques. Cependant, pour la plupart, elles sont encore restreintes aux mots composés qui sont plus stables et moins nombreux. Toutefois, la langue est remplie d’expressions semi-figées qui forment également des unités sémantiques : les expressions adverbiales et les collocations. De même que pour les mots composés traditionnels, l’identification de ces structures limite la complexité combinatoire induite par l’ambiguïté lexicale. Dans cet article, nous détaillons une expérience qui intègre ces notions dans un processus de segmentation en super-chunks, préalable à l’analyse syntaxique. Nous montrons que notre chunker, développé pour le français, atteint une précision et un rappel de 92,9 % et 98,7 %, respectivement. Par ailleurs, les unités polylexicales réalisent 36,6 % des attachements internes aux constituants nominaux et prépositionnels.

La plate-forme logicielle Outilex, qui sera mise à la disposition de la recherche, du développement et de l’industrie, comporte des composants logiciels qui effectuent toutes les opérations fondamentales du traitement automatique du texte écrit : traitements sans lexiques, exploitation de lexiques et de grammaires, gestion de ressources linguistiques. Les données manipulées sont structurées dans des formats XML, et également dans d’autres formats plus compacts, soit lisibles soit binaires, lorsque cela est nécessaire ; les convertisseurs de formats nécessaires sont inclus dans la plate-forme ; les formats de grammaires permettent de combiner des méthodes statistiques avec des méthodes fondées sur des ressources linguistiques. Enfin, des lexiques du français et de l’anglais issus du LADL, construits manuellement et d’une couverture substantielle seront distribués avec la plate-forme sous licence LGPL-LR.

La quantité de documents disponibles via Internet explose. Cette situation nous incite à rechercher de nouveaux outils de localisation d’information dans des documents et, en particulier, à nous pencher sur l’algorithmique des grammaires context-free appliquée à des familles de graphes d’automates finis (strictement finis ou à cycles). Nous envisageons une nouvelle représentation et de nouveaux traitements informatiques sur ces grammaires, afin d’assurer un accès rapide aux données et un stockage peu coûteux en mémoire.