The paper introduces a novel corpus collected in a set of experiments in Italian schools, annotated for the presence of stereotypes, and related categories. It consists of comments written by teenage students in reaction to fabricated fake news, designed to elicit prejudiced responses, by featuring racial stereotypes. We make use of an annotation scheme which takes into account the implicit or explicit nature of different instances of stereotypes, alongside their forms of discredit. We also annotate the stance of the commenter towards the news article, using a schema inspired by rumor and fake news stance detection tasks. Through this rarely studied setting, we provide a preliminary exploration of the production of stereotypes in a more controlled context. Alongside this novel dataset, we provide both quantitative and qualitative analyses of these reactions, to validate the categories used in their annotation. Through this work, we hope to increase the diversity of available data in the study of the propagation and the dynamics of negative stereotypes.

A crucial aspect in abusive language on social media platforms (toxicity, hate speech, harmful stereotypes, etc.) is its inherent contextual nature. In this paper, we focus on the role of conversational context in abusive language detection, one of the most “direct” forms of context in this domain, as given by the conversation threads (e.g., directly preceding message, original post). The incorporation of surrounding messages has proven vital for the accurate human annotation of harmful content. However, many prior works have either ignored this aspect, collecting and processing messages in isolation, or have obtained inconsistent results when attempting to embed such contextual information into traditional classification methods. The reasons behind these findings have not yet been properly addressed. To this end, we propose an analysis of the impact of conversational context in abusive language detection, through: (1) an analysis of prior works and the limitations of the most common concatenation-based approach, which we attempt to address with two alternative architectures; (2) an evaluation of these methods on existing datasets in English, and a new dataset of French tweets annotated for hate speech and stereotypes; and (3) a qualitative analysis showcasing the necessity for context-awareness in ALD, but also its difficulties.

In this paper, we focus on the topics of misinformation and racial hoaxes from a perspective derived from both social psychology and computational linguistics. In particular, we consider the specific case of anti-immigrant feeling as a first case study for addressing racial stereotypes. We describe the first corpus-based study for multilingual racial stereotype identification in social media conversational threads. Our contributions are: (i) a multilingual corpus of racial hoaxes, (ii) a set of common guidelines for the annotation of racial stereotypes in social media texts, and a multi-layered, fine-grained scheme, psychologically grounded on the work by Fiske, including not only stereotype presence, but also contextuality, implicitness, and forms of discredit, (iii) a multilingual dataset in Italian, Spanish, and French annotated following the aforementioned guidelines, and cross-lingual comparative analyses taking into account racial hoaxes and stereotypes in online discussions. The analysis and results show the usefulness of our methodology and resources, shedding light on how racial hoaxes are spread, and enable the identification of negative stereotypes that reinforce them.

Hate speech has unfortunately become a significant phenomenon on social media platforms, and it can cover various topics (misogyny, sexism, racism, xenophobia, etc.) and targets (e.g., black people, women). Various hate speech detection datasets have been proposed, some annotated for specific topics, and others for hateful speech in general. In either case, they often employ different annotation guidelines, which can lead to inconsistencies, even in datasets focusing on the same topics. This can cause issues in models trying to generalize across more data and more topics in order to improve detection accuracy. In this paper, we propose, for the first time, a topic-oriented approach to study generalization across popular hate speech datasets. We first perform a comparative analysis of the performances of Transformer-based models in capturing topic-generic and topic-specific knowledge when trained on different datasets. We then propose a novel, simple yet effective approach to study more precisely which topics are best captured in implicit manifestations of hate, showing that selecting combinations of datasets with better out-of-domain topical coverage improves the reliability of automatic hate speech detection.

De nombreuses tâches sémantiques en TAL font usage de données collectées de manière semiautomatique, ce qui est souvent source d’artefacts indésirables qui peuvent affecter négativement les modèles entraînés sur celles-ci. Avec l’évolution plus récente vers des modèles à usage générique pré-entraînés plus complexes, et moins interprétables, ces biais peuvent conduire à l’intégration de corrélations indésirables dans des applications utilisateurs. Récemment, quelques méthodes ont été proposées pour entraîner des plongements de mots avec une meilleure interprétabilité. Nous proposons une méthode simple qui exploite ces représentations pour détecter de manière préventive des corrélations lexicales faciles à apprendre, dans divers jeux de données. Nous évaluons à cette fin quelques modèles de plongements interprétables populaires pour l’anglais, en utilisant à la fois une évaluation intrinsèque, et un ensemble de tâches sémantiques en aval, et nous utilisons la qualité interprétable des plongements afin de diagnostiquer des biais potentiels dans les jeux de données associés.

Les mesures de similarité textuelle ont une place importante en TAL, du fait de leurs nombreuses applications, en recherche d’information et en classification notamment. En revanche, le dialogue fait moins l’objet d’attention sur cette question. Nous nous intéressons ici à la production d’une similarité dans le contexte d’un corpus de conversations par chat à l’aide de méthodes non-supervisées, exploitant à différents niveaux la notion de sémantique distributionnelle, sous forme d’embeddings. Dans un même temps, pour enrichir la mesure, et permettre une meilleure interprétation des résultats, nous établissons des alignements explicites des tours de parole dans les conversations, en exploitant la distance de Wasserstein, qui permet de prendre en compte leur dimension structurelle. Enfin, nous évaluons notre approche à l’aide d’une tâche externe sur la petite partie annotée du corpus, et observons qu’elle donne de meilleurs résultats qu’une variante plus naïve à base de moyennes.