Misogynistic Content Detection Over Social Networks

The identification of misogynistic content on social networks presents numerous challenges for Natural Language Processing Techniques, primarily due to the intricate nature of its detection, which necessitates a heightened level of contextual awareness. Our research employs a dataset derived from major platforms such as Twitter, Facebook, and Instagram, focusing on posts related to prominent female figures in Italy. The dataset, comprising 16,500 messages, has been meticulously annotated by three independent annotators to discern misogynistic content. Leveraging advanced natural language processing techniques, the study aims to develop an effective model for the automatic identification of misogynistic language within the dynamic context of social media. The findings contribute to the ongoing discourse on mitigating online gender-based harassment, providing valuable insights for the development of robust content moderation mechanisms.