| Resumen |
We present the shared task MiSonGyny organized at IberLEF 2025 as part of the 41st International Congress of the Spanish Society for Natural Language Processing (SEPLN 2025). The objective of this work is to make these problems visible through computational approaches, thereby encouraging the development of natural language processing systems designed for the automatic detection and classification of misogynistic content in Spanish song lyrics. MiSonGyny is composed of two subtasks: Task 1, misogynistic speech detection, detecting song lyrics phrases containing misogynistic language, having two classes, (M) Misogynistic and (NM) Non-Misogynistic, this makes it a binary classification task; Task 2, accurate misogynistic speech detection, this task aims to predict the type of speech present in a song phrase, having three classes, (S) Sexualization, (V) Violence, (H) Hate and (NR) Unrelated. Fifteen teams submitted papers describing their systems. Most approaches implemented Transformer-based architectures to address the task; the best-performing teams incorporated preprocessing strategies, ensembles, and data |
