Servicios Personalizados
Revista
Articulo
Indicadores
- Citado por SciELO
- Accesos
Links relacionados
- Similares en SciELO
Compartir
Computación y Sistemas
versión On-line ISSN 2007-9737versión impresa ISSN 1405-5546
Resumen
CALDERON-SUAREZ, Ricardo; ORTEGA-MENDOZA, Rosa María; MARQUEZ-VERA, Marco Antonio y CASTRO-ESPINOZA, Félix Agustín. Automatic Identification of Misogynistic Content on Social Networks: An Approach based on Knowledge Transfer from Songs. Comp. y Sist. [online]. 2024, vol.28, n.1, pp.283-299. Epub 21-Mar-2024. ISSN 2007-9737. https://doi.org/10.13053/cys-28-1-4896.
This research paper presents a summary of the thesis “Automatic Detection of Misogynistic Content in Social Networks through Knowledge Transfer from Songs”, where the main idea is to leverage the existing knowledge of some songs to transfer linguistic patterns that help to identify manifestations of misogyny in social media. In particular, several learning transfer techniques were analyzed. In addition, a methodology is presented to build, automatically, a collection of songs and another of phrases, both with instances labeled according to the presence or absence of misogynistic content. The major contribution of this research is a data augmentation method that increases the generalization capability of the misogyny detection models by transferring the semantic richness contained in song lyrics. The proposed approach was evaluated in benchmark collections containing texts in Spanish and English, obtaining encouraging results. Compared to robust state-of-the-art approaches, the proposed approach obtained competitive results in English and significant gains in Spanish. This research confirmed the existence of valuable linguistic knowledge in songs, which can be transferred to detect misogynistic content in social media.
Palabras llave : Transfer learning; data augmentation; mysogyny detection; social media.