TaskGen ranked as 6th best team in the TSAR 2025 shared task for English text adaptation to a target CEFR level. Our experiments consisted of prompting a Llama-3.1-8B-Instruct model with linguistic descriptors of the target level, examples of adaptations and multi-step approaches. Our best run, 13th in the overall ranking, applied an ensemble strategy using a voting mechanism to find the most adequate among 10 texts, each produced by a different prompting strategy.

Warning: This paper contains explicit statements of offensive stereotypes which may be upsetting The study of bias, fairness and social impact in Natural Language Processing (NLP) lacks resources in languages other than English. Our objective is to support the evaluation of bias in language models in a multilingual setting. We use stereotypes across nine types of biases to build a corpus containing contrasting sentence pairs, one sentence that presents a stereotype concerning an underadvantaged group and another minimally changed sentence, concerning a matching advantaged group. We build on the French CrowS-Pairs corpus and guidelines to provide translations of the existing material into seven additional languages. In total, we produce 11,139 new sentence pairs that cover stereotypes dealing with nine types of biases in seven cultural contexts. We use the final resource for the evaluation of relevant monolingual and multilingual masked language models. We find that language models in all languages favor sentences that express stereotypes in most bias categories. The process of creating a resource that covers a wide range of language types and cultural settings highlights the difficulty of bias evaluation, in particular comparability across languages and contexts.

Approaches to bias assessment usually require such technical skills that, by design, they leave discrimination experts out. In this paper we present EDIA, a tool that facilitates that experts in discrimination explore social biases in word embeddings and masked language models. Experts can then characterize those biases so that their presence can be assessed more systematically, and actions can be planned to address them. They can work interactively to assess the effects of different characterizations of bias in a given word embedding or language model, which helps to specify informal intuitions in concrete resources for systematic testing.

The expansion of Large Language Models (LLMs) into more serious areas of application, involving decision-making and the forming of public opinion, calls for a more thoughtful treatment of texts. Augmenting them with explicit and understandable argumentative analysis could foster a more reasoned usage of chatbots, text completion mechanisms or other applications. However, it is unclear which aspects of argumentation can be reliably identified and integrated by them. In this paper we propose an adaptation of Wagemans (2016)’s Periodic Table of Arguments to identify different argumentative aspects of texts, with a special focus on hate speech in social media. We have empirically assessed the reliability with which each of these aspects can be automatically identified. We analyze the implications of these results, and how to adapt the proposal to obtain reliable representations of those that cannot be successfully identified.

Since BERT appeared, Transformer language models and transfer learning have become state-of-the-art for natural language processing tasks. Recently, some works geared towards pre-training specially-crafted models for particular domains, such as scientific papers, medical documents, user-generated texts, among others. These domain-specific models have been shown to improve performance significantly in most tasks; however, for languages other than English, such models are not widely available. In this work, we present RoBERTuito, a pre-trained language model for user-generated text in Spanish, trained on over 500 million tweets. Experiments on a benchmark of tasks involving user-generated text showed that RoBERTuito outperformed other pre-trained language models in Spanish. In addition to this, our model has some cross-lingual abilities, achieving top results for English-Spanish tasks of the Linguistic Code-Switching Evaluation benchmark (LinCE) and also competitive performance against monolingual models in English Twitter tasks. To facilitate further research, we make RoBERTuito publicly available at the HuggingFace model hub together with the dataset used to pre-train it.

In this paper, we present a Wikipedia-based approach to develop resources for the legal domain. We establish a mapping between a legal domain ontology, LKIF (Hoekstra et al. 2007), and a Wikipedia-based ontology, YAGO (Suchanek et al. 2007), and through that we populate LKIF. Moreover, we use the mentions of those entities in Wikipedia text to train a specific Named Entity Recognizer and Classifier. We find that this classifier works well in the Wikipedia, but, as could be expected, performance decreases in a corpus of judgments of the European Court of Human Rights. However, this tool will be used as a preprocess for human annotation. We resort to a technique called “curriculum learning” aimed to overcome problems of overfitting by learning increasingly more complex concepts. However, we find that in this particular setting, the method works best by learning from most specific to most general concepts, not the other way round.

The primary aim of the project SENSEM (Sentence Semantics, BFF2003-06456) is the construction of a Lexical Data Base illustrating the syntactic and semantic behavior of each of the senses of the 250 most frequent verbs of Spanish. With this objective in mind, we are currently building an annotated corpus consisting of sentences extracted from the electronic version of the newspaper El Periódico de Catalunya, totalling approximately 1 million words, with 100 examples of each verb. By the time of the conference, we will be about to complete the annotation of 25,000 sentences, which means roughly a corpus of 800,000 words. Approximately 400,000 of them will have been revised. We expect to make the corpus publicly available by the end of 2006.

We present a first approach to the application of a data mining technique, Multiple Sequence Alignment, to the systematization of a polemic aspect of discourse, namely, the expression of contrast, concession, counterargument and semantically similar discursive relations. The representation of the phenomena under study is carried out by very simple techniques, mostly pattern-matching, but the results allow to drive insightful conclusions on the organization of this aspect of discourse: equivalence classes of discourse markers are established, and systematic patterns are discovered, which will be applied in enhancing a discursive parser.