Skip to main content

VERIFICAR Y GARANTIZAR LA CONFIANZA EN LOS LLM EN EL AMBITO BIOANITARIO

AIA2025-163322-C64

•
Funding agency name Agencia Estatal de Investigación
Funding agency acronym AEI
Program Programa Estatal para la Investigación y el Desarrollo Experimental
Subprogram Subprograma Estatal de Generación de Conocimiento Científico-Técnico y Desarrollo Experimental
Call Proyectos de Investigación en el ámbito de la Inteligencia Artificial
Call year 2025
State Plan 2024-2027 Plan Estatal de Investigación Científica y Técnica y de Innovación
Beneficiary centre UNIVERSIDAD DE JAEN

Resultados relacionados

Resultados totales (Incluyendo duplicados): 2
Encontrada(s) 1 página(s)

RUA. Repositorio Institucional de la Universidad de Alicante
Publication Otros (Other)

Nuanced Toxicity Detection in Spanish: A New Corpus and Benchmark Study

  • Mármol-Romero, Alba María
  • Sepúlveda-Torres, Robiert
  • Saquete Boró, Estela
  • Martín Valdivia, María Teresa
  • Ureña López, Luis Alfonso
  • Sepúlveda-Torres, Robiert|||/items/60f24ee2-fe45-4b5c-a70c-c025ed1b5e4a
  • Saquete Boró, Estela|||/items/452393f2-fb8d-4840-b797-4b65a4bc9a87
The rise of toxic content on digital platforms has intensified the demand for automatic moderation tools. While English has benefited from large-scale annotated corpora, Spanish remains under-resourced, particularly for nuanced cases of toxicity such as irony, sarcasm, or indirect aggression. We present an extended version of the NECOS-TOX corpus, comprising 4,011 Spanish comments collected from 16 major news outlets. Each comment is annotated across three levels of toxicity (Non-Toxic, Slightly Toxic, and Toxic), following an iterative annotation protocol that achieved substantial inter-annotator agreement (k = 0.74). To reduce annotation costs while maintaining quality, we employed a human-in-the-loop active learning strategy, with manual correction of model pre-labels. We benchmarked the dataset with traditional machine learning (ML) methods, domain-specific transformers, and instruction-tuned large language models (LLMs). Results show that compact encoder models (e.g., RoBERTa-base-bne, 125M parameters) perform on par with much larger models (e.g., LLaMA-3.1-8B), underscoring the value of in-domain adaptation over raw scale. Our error analysis highlights persistent challenges in distinguishing subtle forms of toxicity, especially sarcasm and implicit insults, and reveals entity-related biases that motivate anonymization strategies. The dataset and trained models are released publicly.
RUA. Repositorio Institucional de la Universidad de Alicante
Publication Otros (Other)

HEART-UJA-UA at PoliticHeadlinES-IberLEF 2026: Generative and Encoder-Based Approaches for Spanish Political Headline Ranking

  • Moreno-Muñoz, Adrián
  • Canal-Esteve, Miquel
  • Martín Valdivia, María Teresa
  • Gutiérrez, Yoan
  • Ureña López, Luis Alfonso
  • Llopis, Fernando
  • Martínez Cámara, Eugenio
  • Canal-Esteve, Miquel|||/items/f1b086bf-bce4-49ea-b765-73081f7a27f9
  • Gutiérrez, Yoan|||/items/8f7f89db-a015-4ed6-85b9-75bd041ee8bd
  • Llopis, Fernando|||/items/e8548380-b70f-44e7-a7fa-c7bed36990d8
This paper presents our participation in PoliticHeadlinES 2026, a shared task on ranking candidate headlines for Spanish political news articles. We compare encoder-based and generative approaches, including direct semantic similarity, a lightweight pairwise neural ranker built on frozen multilingual embeddings, and a retrieval-augmented few-shot prompting pipeline. Our final results show that the generative approach achieves the best overall performance, indicating that in-context ranking with retrieved examples is particularly effective for this task. We also explore multimodal late fusion with CLIP-based image embeddings and different strategies for representing long articles, finding that visual information provides limited benefits and that truncating the article to the first 512 tokens remains a strong baseline for encoder models. Overall, the HEART-UJA-UA team ranked sixth out of 20 participants in the official leaderboard, obtaining PA-nDCG scores of 0.885 in Task 1 and 0.896 in Task 2.