Classified "A" by ANVUR in the fields 11/D1, 11/D2 Scientific in the field 14.
Abstract:
The article examines the integration of machine learning techniques and socio-educational analysis to classify online hate speech targeting Pope Francis, focusing on Twitter and Telegram from January 2022 to April 2024. The aim is to understand how hate is articulated concerning specific themes addressed by the pontificate, employing the "Online Hate Spectrum" model, previously applied to other forms of hate speech such as anti-Romani sentiment, Islamophobia, and antisemitism. On Twitter, the analysis was conducted across eight thematic categories, selecting tweets using machine learning techniques based on keywords and research criteria defined by the Mediavox Observatory of the Catholic University. After identifying monthly peaks in hate speech, a manual classification determined the presence of discriminatory content, analyzing the rhetoric used and specific targets. In parallel, the study on Telegram examined 92 channels disseminating hostile and conspiratorial narratives, mapping content, interactions, and the use of images. The analysis revealed a centralized network of shares and comments, characterized by an informational bubble with few origin points. Channels were classified based on their tone, distinguishing between neutral dissemination, constructive criticism, and provocative attitudes. The results provide a detailed picture of online hate dynamics against Pope Francis, highlighting the association between conversational flows and processes of target selection, and contribute to a better understanding of the strategies needed to effectively counter online hate speech.
