Explaining why content is hateful using natural language is crucial for fostering transparency in automated content moderation systems. However, evaluating the quality of such explanations remains an open challenge. General-purpose reward models (RMs), commonly used for scoring natural language outputs, are typically optimized for broad notions of safety. We argue that this optimization penalizes situations where references to stereotypes or offensive content are essential for explanations with higher explanatory fidelity. To address this gap, we introduce SBIC-Explain, a human-validated dataset of 370,788 LLM generated NLEs for offensive content, spanning three levels of human-annotated contextual richness: Tier 1: text-only, Tier 2: + classification-aware, and Tier 3: + semantics-informed. We hypothesize that as human-annotated context increases, explanations should lead to higher perceived explanations with higher explanatory fidelity. Yet, we find that existing RMs systematically assign lower scores to more contextually rich (and often more offensive) explanations, revealing a misalignment between model preferences and explanatory fidelity for this context. We propose HARM (Hate-Aware Reward Model), a RM that integrates interpretable signals to better align reward scores with the needs of hate speech explanation. HARM outperforms general-purpose baselines, improving NLE pair-wise preference. Available at: https://github.com/Lorenzo815/HARM.

This study introduces the system submitted to the SemEval 2022 Task 11: MultiCoNER (Multilingual Complex Named Entity Recognition) by the UC3M-PUCPR team. We proposed an ensemble of transformer-based models for entity recognition in cross-domain texts. Our deep learning method benefits from the transformer architecture, which adopts the attention mechanism to handle the long-range dependencies of the input text. Also, the ensemble approach for named entity recognition (NER) improved the results over baselines based on individual models on two of the three tracks we participated in. The ensemble model for the code-mixed task achieves an overall performance of 76.36% F1-score, a 2.85 percentage point increase upon our individually best model for this task, XLM-RoBERTa-large (73.51%), outperforming the baseline provided for the shared task by 18.26 points. Our preliminary results suggest that contextualized language models ensembles can, even if modestly, improve the results in extracting information from unstructured data.

With the growing number of electronic health record data, clinical NLP tasks have become increasingly relevant to unlock valuable information from unstructured clinical text. Although the performance of downstream NLP tasks, such as named-entity recognition (NER), in English corpus has recently improved by contextualised language models, less research is available for clinical texts in low resource languages. Our goal is to assess a deep contextual embedding model for Portuguese, so called BioBERTpt, to support clinical and biomedical NER. We transfer learned information encoded in a multilingual-BERT model to a corpora of clinical narratives and biomedical-scientific papers in Brazilian Portuguese. To evaluate the performance of BioBERTpt, we ran NER experiments on two annotated corpora containing clinical narratives and compared the results with existing BERT models. Our in-domain model outperformed the baseline model in F1-score by 2.72%, achieving higher performance in 11 out of 13 assessed entities. We demonstrate that enriching contextual embedding models with domain literature can play an important role in improving performance for specific NLP tasks. The transfer learning process enhanced the Portuguese biomedical NER model by reducing the necessity of labeled data and the demand for retraining a whole new model.