LGTBfobia en redes sociales: Revisión sistemática de la detección y clasificación de discurso de odio a gran escala

Abstract

Hate speech is a significant threat to social diversity and human rights. Detecting anti-LGBT hate speech on social media is a complex area of research that requires a multifaceted and adaptable approach. This work aims to present a review of recent efforts in the investigation of hate speech towards the LGBT population on social networks and explore the use of computational techniques for the detection and classification of these speeches. Methodologies for detecting LGBTphobic hate speech on social networks are examined, covering theoretical and practical aspects, analyzing recent resources, discussing classification techniques and methods, and identifying challenges. The time frame of this literature review focuses between January 1, 2019 to February 29, 2024, including works with a specific focus on hate speech directed at the LGBT community through digital networks, emphasizing studies that apply computational detection methods. In the search process we have reviewed documents in different databases, such as Scopus and WoS. We identified 28 relevant works in academic databases, noting that it is essential to take into account linguistic and cultural diversity, as well as the sensitivity of the topic when collecting data and developing detection models. Much of the previous research has focused on using computational methods to detect large amounts of hate speech on networks and creating strategies to combat it. Furthermore, there is a trend towards the use of deep learning models, such as BERT, and the need to adapt these models to specific contexts to improve their effectiveness is emphasized. Some emerging trends centered on hopeful discourse, counternarratives, and assessment of the level of toxicity are also mentioned. As future lines, the exploration of multilingual approaches and interdisciplinary collaboration to address the challenges in the detection and prevention of hate speech on social networks stands out.