Enhancing the Detection of Sexist Messages through a Multi-Profile-Based Ensemble Approach

• Autores: Martha Paola Jiménez-Martínez, Irvin Hussein Lopez Nava, Manuel Montes Gómez
• Localización: Computación y Sistemas (CyS), ISSN 1405-5546, ISSN-e 2007-9737, Vol. 29, Nº. 1, 2025, págs. 283-294
• Idioma: inglés
• Enlaces
  • Texto completo
• Resumen

  • Abstract: Sexism in language perpetuates harmful stereotypes, especially in cultures with deeply ingrained traditional gender roles, such as Mexico. While detection of misogynistic content in English has advanced, detecting sexist language in Spanish is less explored. This study uses the EXIST corpus, annotated by various demographic groups, to examine differing perceptions of sexism across genders and ages. Our analysis finds significant perception discrepancies, with 25% of texts showing disagreements between male and female annotators. We propose an ensemble classification model that integrates outputs from gender-specific and age-specific models based on ROBERTuito, achieving an F1 score of 0.854. To gain insights into our best classifier’s decision-making, we present an error analysis based on the visualization of attention weights, which helps us identify the most relevant words in the detection of subtle sexism. Additionally, we leverage ChatGPT’s capabilities to model language nuances, generating potential interpretations of texts associated with the classifications provided by our approach. This study underscores the importance of demographic considerations in sexist language detection and demonstrates that combining diverse perspectives with advanced techniques can enhance detection in Spanish social media.