Identification of sexism on social media

Abstract

[en] With the rapid advancement of communication technology, smartphone usage, and sophisticated algorithms, social media has become an integral and inseparable part of modern society. Consequently, the prevalence of sexist content on these platforms has emerged as a significant and far-reaching issue. This form of online harassment not only perpetuates gender inequalities but also poses significant psychological and emotional harm to individuals targeted by such content. Thus, it is imperative to address this problem and take proactive measures to mitigate its impact. The main goal of this work is to study, identify and analyze the process of detection of sexist content through the application of natural language processing techniques. The study utilizes two datasets from the EXIST competition, a shared task of sEXism Identification in Social neTworks from IberLeF 2021 and CLEF 2023. Five state-of-the-art language models, based on Transformers and Deep Learning, are trained and validated for subsequent comparison. The primary objective is to identify instances of online sexism and determine the optimal framework for each task, which accurately reflects real-world scenarios.