Extração de características para identificação de discurso de ódio em documentos

Abstract

Social media is increasingly present in people’s lives, including tools that allow users to collaborate with the creation of the content exposed in it. Many users use this functionality to post texts spreading illicit or criminal content. This offensive content need be removed soon as possible otherwise more and more people we see and can propagate through the internet, reaching a more significant number of victims encouraging the occurrence of other crimes. Thisworkproposestheextractionofcharacteristicsfromtextusingnaturallanguageprocessing techniques and machine learning to detect hate speech automatically. This type of hate crime, in general, is focused on the most vulnerable groups in society, and the harmful effects can lead to increased social exclusion and violence against such groups.