The paper describes a dataset composed of two sub-corpora from two different sources in Italian. The QUEEREOTYPES corpus includes social media texts regarding LGBTQIA+ individuals, behaviors, ideology and events. The texts were collected from Facebook and Twitter in 2018 and were annotated for the presence of stereotypes, and orthogonal dimensions (such as hate speech, aggressiveness, offensiveness, and irony in one sub-corpus, and stance in the other). The resource was developed by Natural Language Processing researchers together with activists from an Italian LGBTQIA+ not-for-profit organization. The creation of the dataset allows the NLP community to study stereotypes against marginalized groups, individuals and, ultimately, to develop proper tools and measures to reduce the online spread of such stereotypes. A test for the robustness of the language resource has been performed by means of 5-fold cross-validation experiments. Finally, text classification experiments have been carried out with a fine-tuned version of AlBERTo (a BERT-based model pre-trained on Italian tweets) and mBERT, obtaining good results on the task of stereotype detection, suggesting that stereotypes towards different targets might share common traits.

Supervised classification heavily depends on datasets annotated by humans. However, in subjective tasks such as toxicity classification, these annotations often exhibit low agreement among raters. Annotations have commonly been aggregated by employing methods like majority voting to determine a single ground truth label. In subjective tasks, aggregating labels will result in biased labeling and, consequently, biased models that can overlook minority opinions. Previous studies have shed light on the pitfalls of label aggregation and have introduced a handful of practical approaches to tackle this issue. Recently proposed multi-annotator models, which predict labels individually per annotator, are vulnerable to under-determination for annotators with few samples. This problem is exacerbated in crowdsourced datasets. In this work, we propose Annotator Aware Representations for Texts (AART) for subjective classification tasks. Our approach involves learning representations of annotators, allowing for exploration of annotation behaviors. We show the improvement of our method on metrics that assess the performance on capturing individual annotators’ perspectives. Additionally, we demonstrate fairness metrics to evaluate our model’s equability of performance for marginalized annotators compared to others.

NLP datasets annotated with human judgments are rife with disagreements between the judges. This is especially true for tasks depending on subjective judgments such as sentiment analysis or offensive language detection. Particularly in these latter cases, the NLP community has come to realize that the common approach of reconciling’ these different subjective interpretations risks misrepresenting the evidence. Many NLP researchers have therefore concluded that rather than eliminating disagreements from annotated corpora, we should preserve themindeed, some argue that corpora should aim to preserve all interpretations produced by annotators. But this approach to corpus creation for NLP has not yet been widely accepted. The objective of the Le-Wi-Di series of shared tasks is to promote this approach to developing NLP models by providing a unified framework for training and evaluating with such datasets. We report on the second such shared task, which differs from the first edition in three crucial respects: (i) it focuses entirely on NLP, instead of both NLP and computer vision tasks in its first edition; (ii) it focuses on subjective tasks, instead of covering different types of disagreements as training with aggregated labels for subjective NLP tasks is in effect a misrepresentation of the data; and (iii) for the evaluation, we concentrated on soft approaches to evaluation. This second edition of Le-Wi-Di attracted a wide array of partici- pants resulting in 13 shared task submission papers.

We present EPIC (English Perspectivist Irony Corpus), the first annotated corpus for irony analysis based on the principles of data perspectivism. The corpus contains short conversations from social media in five regional varieties of English, and it is annotated by contributors from five countries corresponding to those varieties. We analyse the resource along the perspectives induced by the diversity of the annotators, in terms of origin, age, and gender, and the relationship between these dimensions, irony, and the topics of conversation. We validate EPIC by creating perspective-aware models that encode the perspectives of annotators grouped according to their demographic characteristics. Firstly, the performance of perspectivist models confirms that different annotators induce very different models. Secondly, in the classification of ironic and non-ironic texts, perspectivist models prove to be generally more confident than the non-perspectivist ones. Furthermore, comparing the performance on a perspective-based test set with those achieved on a gold standard test set, we can observe how perspectivist models tend to detect more precisely the positive class, showing their ability to capture the different perceptions of irony. Thanks to these models, we are moreover able to show interesting insights about the variation in the perception of irony by the different groups of annotators, such as among different generations and nationalities.

This paper introduces the Unified Interactive Natural Understanding of the Italian Language (UINAUIL), a benchmark of six tasks for Italian Natural Language Understanding. We present a description of the tasks and software library that collects the data from the European Language Grid, harmonizes the data format, and exposes functionalities to facilitates data manipulation and the evaluation of custom models. We also present the results of tests conducted with available Italian and multilingual language models on UINAUIL, providing an updated picture of the current state of the art in Italian NLU.

Research in the field of NLP has recently focused on the variability that people show in selecting labels when performing an annotation task. Exploiting disagreements in annotations has been shown to offer advantages for accurate modelling and fair evaluation. In this paper, we propose a strongly perspectivist model for supervised classification of natural language utterances. Our approach combines the predictions of several perspective-aware models using key information of their individual confidence to capture the subjectivity encoded in the annotation of linguistic phenomena. We validate our method through experiments on two case studies, irony and hate speech detection, in in-domain and cross-domain settings. The results show that confidence-based ensembling of perspective-aware models seems beneficial for classification performance in all scenarios. In addition, we demonstrate the effectiveness of our method with automatically extracted perspectives from annotations when the annotators’ metadata are not available.

The European Language Grid enables researchers and practitioners to easily distribute and use NLP resources and models, such as corpora and classifiers. We describe in this paper how, during the course of our EVALITA4ELG project, we have integrated datasets and systems for the Italian language. We show how easy it is to use the integrated systems, and demonstrate in case studies how seamless the application of the platform is, providing Italian NLP for everyone.

Despite the large number of computational resources for emotion recognition, there is a lack of data sets relying on appraisal models. According to Appraisal theories, emotions are the outcome of a multi-dimensional evaluation of events. In this paper, we present APPReddit, the first corpus of non-experimental data annotated according to this theory. After describing its development, we compare our resource with enISEAR, a corpus of events created in an experimental setting and annotated for appraisal. Results show that the two corpora can be mapped notwithstanding different typologies of data and annotations schemes. A SVM model trained on APPReddit predicts four appraisal dimensions without significant loss. Merging both corpora in a single training set increases the prediction of 3 out of 4 dimensions. Such findings pave the way to a better performing classification model for appraisal prediction.

This work describes the process of creating a corpus of Twitter conversations annotated for the presence of counterspeech in response to toxic speech related to axes of discrimination linked to sexism, racism and homophobia. The main novelty is an annotated dataset comprising relevant tweets in their context of occurrence. The corpus is made up of tweets and responses captured by different profiles replying to discriminatory content or objectionably couched news. An annotation scheme was created to make explicit the knowledge on the dimensions of toxic speech and counterspeech.An analysis of the collected and annotated data and of the IAA that emerged during the annotation process is included. Moreover, we report about preliminary experiments on automatic counterspeech detection, based on supervised automatic learning models trained on the new dataset. The results highlight the fundamental role played by the context in this detection task, confirming our intuitions about the importance to collect tweets in their context of occurrence.

Hate speech recognizers may mislabel sentences by not considering the different opinions that society has on selected topics. In this paper, we show how explainable machine learning models based on syntax can help to understand the motivations that induce a sentence to be offensive to a certain demographic group. By comparing and contrasting the results, we show the key points that make a sentence labeled as hate speech and how this varies across different ethnic groups.

Inside the NLP community there is a considerable amount of language resources created, annotated and released every day with the aim of studying specific linguistic phenomena. Despite a variety of attempts in order to organize such resources has been carried on, a lack of systematic methods and of possible interoperability between resources are still present. Furthermore, when storing linguistic information, still nowadays, the most common practice is the concept of “gold standard”, which is in contrast with recent trends in NLP that aim at stressing the importance of different subjectivities and points of view when training machine learning and deep learning methods. In this paper we present O-Dang!: The Ontology of Dangerous Speech Messages, a systematic and interoperable Knowledge Graph (KG) for the collection of linguistic annotated data. O-Dang! is designed to gather and organize Italian datasets into a structured KG, according to the principles shared within the Linguistic Linked Open Data community. The ontology has also been designed to account a perspectivist approach, since it provides a model for encoding both gold standard and single-annotator labels in the KG. The paper is structured as follows. In Section 1 the motivations of our work are outlined. Section 2 describes the O-Dang! Ontology, that provides a common semantic model for the integration of datasets in the KG. The Ontology Population stage with information about corpora, users, and annotations is presented in Section 3. Finally, in Section 4 an analysis of offensiveness across corpora is provided as a first case study for the resource.

We introduce HateBERT, a re-trained BERT model for abusive language detection in English. The model was trained on RAL-E, a large-scale dataset of Reddit comments in English from communities banned for being offensive, abusive, or hateful that we have curated and made available to the public. We present the results of a detailed comparison between a general pre-trained language model and the retrained version on three English datasets for offensive, abusive language and hate speech detection tasks. In all datasets, HateBERT outperforms the corresponding general BERT model. We also discuss a battery of experiments comparing the portability of the fine-tuned models across the datasets, suggesting that portability is affected by compatibility of the annotated phenomena.

Best-worst Scaling (BWS) is a methodology for annotation based on comparing and ranking instances, rather than classifying or scoring individual instances. Studies have shown the efficacy of this methodology applied to NLP tasks in terms of a higher quality of the datasets produced by following it. In this system demonstration paper, we present Litescale, a free software library to create and manage BWS annotation tasks. Litescale computes the tuples to annotate, manages the users and the annotation process, and creates the final gold standard. The functionalities of Litescale can be accessed programmatically through a Python module, or via two alternative user interfaces, a textual console-based one and a graphical Web-based one. We further developed and deployed a fully online version of Litescale complete with multi-user support.

Evaluation is of paramount importance in data-driven research fields such as Natural Language Processing (NLP) and Computer Vision (CV). Current evaluation practice largely hinges on the existence of a single “ground truth” against which we can meaningfully compare the prediction of a model. However, this comparison is flawed for two reasons. 1) In many cases, more than one answer is correct. 2) Even where there is a single answer, disagreement among annotators is ubiquitous, making it difficult to decide on a gold standard. We argue that the current methods of adjudication, agreement, and evaluation need serious reconsideration. Some researchers now propose to minimize disagreement and to fix datasets. We argue that this is a gross oversimplification, and likely to conceal the underlying complexity. Instead, we suggest that we need to better capture the sources of disagreement to improve today’s evaluation practice. We discuss three sources of disagreement: from the annotator, the data, and the context, and show how this affects even seemingly objective tasks. Datasets with multiple annotations are becoming more common, as are methods to integrate disagreement into modeling. The logical next step is to extend this to evaluation.

Abusive language detection is an unsolved and challenging problem for the NLP community. Recent literature suggests various approaches to distinguish between different language phenomena (e.g., hate speech vs. cyberbullying vs. offensive language) and factors (degree of explicitness and target) that may help to classify different abusive language phenomena. There are data sets that annotate the target of abusive messages (i.e.OLID/OffensEval (Zampieri et al., 2019a)). However, there is a lack of data sets that take into account the degree of explicitness. In this paper, we propose annotation guidelines to distinguish between explicit and implicit abuse in English and apply them to OLID/OffensEval. The outcome is a newly created resource, AbuseEval v1.0, which aims to address some of the existing issues in the annotation of offensive and abusive language (e.g., explicitness of the message, presence of a target, need of context, and interaction across different phenomena).

Swearing plays an ubiquitous role in everyday conversations among humans, both in oral and textual communication, and occurs frequently in social media texts, typically featured by informal language and spontaneous writing. Such occurrences can be linked to an abusive context, when they contribute to the expression of hatred and to the abusive effect, causing harm and offense. However, swearing is multifaceted and is often used in casual contexts, also with positive social functions. In this study, we explore the phenomenon of swearing in Twitter conversations, taking the possibility of predicting the abusiveness of a swear word in a tweet context as the main investigation perspective. We developed the Twitter English corpus SWAD (Swear Words Abusiveness Dataset), where abusive swearing is manually annotated at the word level. Our collection consists of 1,511 unique swear words from 1,320 tweets. We developed models to automatically predict abusive swearing, to provide an intrinsic evaluation of SWAD and confirm the robustness of the resource. We also present the results of a glass box ablation study in order to investigate which lexical, syntactic, and affective features are more informative towards the automatic prediction of the function of swearing.

This paper describes our participation to the TRAC-2 Shared Tasks on Aggression Identification. Our team, FlorUniTo, investigated the applicability of using an abusive lexicon to enhance word embeddings towards improving detection of aggressive language. The embeddings used in our paper are word-aligned pre-trained vectors for English, Hindi, and Bengali, to reflect the languages in the shared task data sets. The embeddings are retrofitted to a multilingual abusive lexicon, HurtLex. We experimented with an LSTM model using the original as well as the transformed embeddings and different language and setting variations. Overall, our systems placed toward the middle of the official rankings based on weighted F1 score. However, the results on the development and test sets show promising improvements across languages, especially on the misogynistic aggression sub-task.

This paper presents an in-depth investigation of the effectiveness of dependency-based syntactic features on the irony detection task in a multilingual perspective (English, Spanish, French and Italian). It focuses on the contribution from syntactic knowledge, exploiting linguistic resources where syntax is annotated according to the Universal Dependencies scheme. Three distinct experimental settings are provided. In the first, a variety of syntactic dependency-based features combined with classical machine learning classifiers are explored. In the second scenario, two well-known types of word embeddings are trained on parsed data and tested against gold standard datasets. In the third setting, dependency-based syntactic features are combined into the Multilingual BERT architecture. The results suggest that fine-grained dependency-based syntactic information is informative for the detection of irony.

We introduce an approach to multilingual Offensive Language Detection based on the mBERT transformer model. We download extra training data from Twitter in English, Danish, and Turkish, and use it to re-train the model. We then fine-tuned the model on the provided training data and, in some configurations, implement transfer learning approach exploiting the typological relatedness between English and Danish. Our systems obtained good results across the three languages (.9036 for EN, .7619 for DA, and .7789 for TR).

The detection of abusive or offensive remarks in social texts has received significant attention in research. In several related shared tasks, BERT has been shown to be the state-of-the-art. In this paper, we propose to utilize lexical features derived from a hate lexicon towards improving the performance of BERT in such tasks. We explore different ways to utilize the lexical features in the form of lexicon-based encodings at the sentence level or embeddings at the word level. We provide an extensive dataset evaluation that addresses in-domain as well as cross-domain detection of abusive content to render a complete picture. Our results indicate that our proposed models combining BERT with lexical features help improve over a baseline BERT model in many of our in-domain and cross-domain experiments.

We describe the system presented at the SR’19 shared task by the DipInfoUnito team. Our approach is based on supervised machine learning. In particular, we divide the SR task into two independent subtasks, namely word order prediction and morphology inflection prediction. Two neural networks with different architectures run on the same input structure, each producing a partial output which is recombined in the final step in order to produce the predicted surface form. This work is a direct successor of the architecture presented at SR’19.

The paper describes the organization of the SemEval 2019 Task 5 about the detection of hate speech against immigrants and women in Spanish and English messages extracted from Twitter. The task is organized in two related classification subtasks: a main binary subtask for detecting the presence of hate speech, and a finer-grained one devoted to identifying further features in hateful contents such as the aggressive attitude and the target harassed, to distinguish if the incitement is against an individual rather than a group. HatEval has been one of the most popular tasks in SemEval-2019 with a total of 108 submitted runs for Subtask A and 70 runs for Subtask B, from a total of 74 different teams. Data provided for the task are described by showing how they have been collected and annotated. Moreover, the paper provides an analysis and discussion about the participant systems and the results they achieved in both subtasks.

This paper describes the system developed by the DipInfo-UniTo team to participate to the shallow track of the Surface Realization Shared Task 2018. The system employs two separate neural networks with different architectures to predict the word ordering and the morphological inflection independently from each other. The UniTO realizer is language independent, and its simple architecture allowed it to be scored in the central part of the final ranking of the shared task.

This paper describes the results of the first Shared Task on Multilingual Emoji Prediction, organized as part of SemEval 2018. Given the text of a tweet, the task consists of predicting the most likely emoji to be used along such tweet. Two subtasks were proposed, one for English and one for Spanish, and participants were allowed to submit a system run to one or both subtasks. In total, 49 teams participated to the English subtask and 22 teams submitted a system run to the Spanish subtask. Evaluation was carried out emoji-wise, and the final ranking was based on macro F-Score. Data and further information about this task can be found at https://competitions.codalab.org/competitions/17344.

Frame semantics is a well-established framework to represent the meaning of natural language in computational terms. In this work, we aim to propose a quantitative measure of relatedness between pairs of frame instances. We test our method on a dataset of sentence pairs, highlighting the correlation between our metric and human judgments of semantic similarity. Furthermore, we propose an application of our measure for clustering frame instances to extract prototypical knowledge from natural language.

What would be a good method to provide a large collection of semantically annotated texts with formal, deep semantics rather than shallow? We argue that a bootstrapping approach comprising state-of-the-art NLP tools for parsing and semantic interpretation, in combination with a wiki-like interface for collaborative annotation of experts, and a game with a purpose for crowdsourcing, are the starting ingredients for fulfilling this enterprise. The result is a semantic resource that anyone can edit and that integrates various phenomena, including predicate-argument structure, scope, tense, thematic roles, rhetorical relations and presuppositions, into a single semantic formalism: Discourse Representation Theory. Taking texts rather than sentences as the units of annotation results in deep semantic representations that incorporate discourse structure and dependencies. To manage the various (possibly conflicting) annotations provided by experts and non-experts, we introduce a method that stores ``Bits of Wisdom'' in a database as stand-off annotations.