Hate speech research between algorithmic and qualitative evaluations.
A case study of hate speech on Twitter
Pasta, S.
At an international level, the field of Hate Studies is marked by a significant amount of research aimed at automating detection processes and creating an algorithm capable of detecting online hatred. This contribution presents an analysis of the Mediavox Observatory, which combines a socio-educational approach and automated computer processing. This case deals with the classification of anti-Gypsy hate speech on Twitter. The methodology used is part of social network analysis (SNA) techniques. Data were collected using the open-source Python library GetOldTweets3 and the corpus was manually classified by experts in the field (“annotators”). The last step is to subject the results to a confusion matrix, i.e. a tool to analyse the errors made by a machine learning model. The case study shows that there is a need to continue experimenting with research that integrates the two steps of human and automatic classification.
Stefano Pasta
Stefano Pasta is a researcher in Didactics and Special Pedagogy at the Faculty of Education of the Catholic University of Milan, where he teaches Methodology of educational and special activities and holds training modules at the University’s Advanced Training and Master courses. He is a member of CREMIT and carries out research on online hate within the Mediavox Observatory.