Natural Language Processing and Generation systems have recently shown the potential to complement and streamline the costly and time-consuming job of professional fact-checkers. In this work, we lift several constraints of current state-of-the-art pipelines for automated fact-checking based on the Retrieval-Augmented Generation (RAG) paradigm. Our goal is to benchmark, following professional fact-checking practices, RAG-based methods for the generation of verdicts - i.e., short texts discussing the veracity of a claim - evaluating them on stylistically complex claims and heterogeneous, yet reliable, knowledge bases. Our findings show a complex landscape, where, for example, LLM-based retrievers outperform other retrieval techniques, though they still struggle with heterogeneous knowledge bases; larger models excel in verdict faithfulness, while smaller models provide better context adherence, with human evaluations favouring zero-shot and one-shot approaches for informativeness, and fine-tuned models for emotional alignment.

In this work we present a system for multilingual olfactory information extraction covering six European languages, introducing new models to extract olfactory information from large amounts of text in a structured and scalable way. For the task we rely on a supervised multi-task approach to detect olfactory related text adopting a FrameNet-like structure, identifying the lexical units triggering the smell event and a related set of frame elements.

Olfactory references play a crucial role in our memory and, more generally, in our experiences, since researchers have shown that smell is the sense that is most directly connected with emotions. Nevertheless, only few works in NLP have tried to capture this sensory dimension from a computational perspective. One of the main challenges is the lack of a systematic and consistent taxonomy of olfactory information, where concepts are organised also in a multi-lingual perspective. WordNet represents a valuable starting point in this direction, which can be semi-automatically extended taking advantage of Google n-grams and of existing language models. In this work we describe the process that has led to the semi-automatic development of a taxonomy for olfactory information in four languages (English, French, German and Italian), detailing the different steps and the intermediate evaluations. Along with being multi-lingual, the taxonomy also encloses temporal marks for olfactory terms thus making it a valuable resource for historical content analysis. The resource has been released and is freely available.

Since state-of-the-art approaches to offensive language detection rely on supervised learning, it is crucial to quickly adapt them to the continuously evolving scenario of social media.While several approaches have been proposed to tackle the problem from an algorithmic perspective, so to reduce the need for annotated data, less attention has been paid to the quality of these data.Following a trend that has emerged recently, we focus on the level of agreement among annotators while selecting data to create offensive language datasets, a task involving a high level of subjectivity.Our study comprises the creation of three novel datasets of English tweets covering different topics and having five crowd-sourced judgments each.We also present an extensive set of experiments showing that selecting training and test data according to different levels of annotators' agreement has a strong effect on classifiers performance and robustness.Our findings are further validated in cross-domain experiments and studied using a popular benchmark dataset.We show that such hard cases, where low agreement is present, are not necessarily due to poor-quality annotation and we advocate for a higher presence of ambiguous cases in future datasets, particularly in test sets, to better account for the different points of view expressed online.

Automated complex word identification (CWI) is a crucial task in several applications, from readability assessment to lexical simplification. So far, several works have modeled CWI with the goal of targeting the needs of non-native speakers. However, studies in language acquisition show that different native languages can create positive or negative interferences w.r.t. reading comprehension, favouring or hindering the understanding of a document in a foreign language. Therefore, we propose to modify CWI to address the specific difficulties connected to different native languages. In particular, we present a pipeline that, based on the user native language, identifies complex terms by automatically detecting cognates and false friends on the fly. The selection presented by the CWI module is adaptive in that it changes depending on the native language of the user. We implement and evaluate our approach for four different native languages (French, English, German and Spanish), in a setting where documents are written in Italian and should be read by language learners with low proficiency. We show that a personalised strategy based on false friend detection identifies complex terms that are different from those usually selected with standard approaches based on word frequency.

The increasing popularity of social media platforms such as Twitter and Facebook has led to a rise in the presence of hate and aggressive speech on these platforms. Despite the number of approaches recently proposed in the Natural Language Processing research area for detecting these forms of abusive language, the issue of identifying hate speech at scale is still an unsolved problem. In this article, we propose a robust neural architecture that is shown to perform in a satisfactory way across different languages; namely, English, Italian, and German. We address an extensive analysis of the obtained experimental results over the three languages to gain a better understanding of the contribution of the different components employed in the system, both from the architecture point of view (i.e., Long Short Term Memory, Gated Recurrent Unit, and bidirectional Long Short Term Memory) and from the feature selection point of view (i.e., ngrams, social network–specific features, emotion lexica, emojis, word embeddings). To address such in-depth analysis, we use three freely available datasets for hate speech detection on social media in English, Italian, and German.

In this work, we apply argumentation mining techniques, in particular relation prediction, to study political speeches in monological form, where there is no direct interaction between opponents. We argue that this kind of technique can effectively support researchers in history, social and political sciences, which must deal with an increasing amount of data in digital form and need ways to automatically extract and analyse argumentation patterns. We test and discuss our approach based on the analysis of documents issued by R. Nixon and J. F. Kennedy during 1960 presidential campaign. We rely on a supervised classifier to predict argument relations (i.e., support and attack), obtaining an accuracy of 0.72 on a dataset of 1,462 argument pairs. The application of argument mining to such data allows not only to highlight the main points of agreement and disagreement between the candidates' arguments over the campaign issues such as Cuba, disarmament and health-care, but also an in-depth argumentative analysis of the respective viewpoints on these topics.

We present a topic-based analysis of agreement and disagreement in political manifestos, which relies on a new method for topic detection based on key concept clustering.Our approach outperforms both standard techniques like LDA and a state-of-the-art graph-based method, and provides promising initial results for this new task in computational social science.

The automated comparison of points of view between two politicians is a very challenging task, due not only to the lack of annotated resources, but also to the different dimensions participating to the definition of agreement and disagreement. In order to shed light on this complex task, we first carry out a pilot study to manually annotate the components involved in detecting agreement and disagreement. Then, based on these findings, we implement different features to capture them automatically via supervised classification. We do not focus on debates in dialogical form, but we rather consider sets of documents, in which politicians may express their position with respect to different topics in an implicit or explicit way, like during an electoral campaign. We create and make available three different datasets.