RECONHECIMENTO AUTOMÁTICO DE NARRATIVAS VIOLENTAS CONTRA A MULHER UTILIZANDO CLASSIFICADORES NEURAIS

Abstract

Automatically identifying traces of violence against women in social media is of utmost importance, given their status in modern human relationships. This article aims to implement a neural classifier capable of identifying abusive speeches in tweets. The creation of a classifier covers the steps of creating a dictionary of words related to the theme; the creation of a database through the collection of Twitter posts; the use of text pre-processing techniques and natural language processing and the training of an Artificial Neural Network. The results of the classification process will be presented based on the metrics accuracy, precision, F1 score and recall. It is expected that the classifier will be able to be used in order to combat the practice of violence against women in a efficient and fast way, in addition to allowing social networks to develop automatic methods to block and/or ban posts of this nature.