SocialHaterBERT: automatic detection and monitoring of hate speech on Twitter through a dichotomous approach based on textual analysis and user profiles

Social media are real-world sensors that can be used to gauge a society's pulse. The vast unfiltered flood of messages is an alarming phenomenon in society, especially if when containing hate speech. In this project, we present SocialHaterBERT, an intelligent system currently being used by the Spanish National Office Against Hate Crimes that identifies and monitors the evolution of hate speech in Twitter. The contributions of this research are many-fold: (1) It introduces the first intelligent system that monitors and visualizes hate speech in social media using social network analysis techniques. (2) It introduces an algorithm that examines features other than those found in the text in an innovative way. Experiments on a case study demonstrate its utility in identifying senders and receivers of hate messages (i.e. bullying) within a high-school class, as well as monitoring group dynamics and toxic peaks in a visual and intuitive manner for authorities.

The project started on 01/01/2018 and is still Running.