Uma abordagem para extrair relatos de agressões contra mulheres no Twitter e enquadrar na Lei Maria da Penha

Abstract

Gender violence is a repeated violence that causes both physical and psychological damages in a fair amount of women around the world. In Brazil the Maria da Penha law through its 15 years of existence remains a national symbol in the fight against gender violence and accumulates millions of lawsuits at court. The COVID-19 pandemic increased considerably the number of violent cases in social isolation and there is a lack of support projects for this group. As an alternative channel to traditional media, the social networks gave women voice in this search for justice and created various research groups who are approaching possible law enforcement from innovative perspectives. For instance, literature in this field highlights key concepts as technopolitics, cyberfeminism, cyberspace and also brings into discussion the theory about the affect of social networks. Other projects analyze online movements such as #MeToo movement, “Não mereço ser estuprada” and dozens of other hashtags that went viral on social media and study ways about how to classify these reports. The goal of our project is to investigate if the reporting of aggression posted on social networks can be covered by Maria da Penha law and to analyze the frequency which type of violence described in the Maria da Penha law. So we developed a methodology to collect posts from social networks and classify them according to the five types of domestic violence covered by Maria da Penha law, i.e., Moral, Sexual, Psychological, Physical, and Property. We validated this methodology by implementing a working prototype that automates the overall proposed approach and takes advantage of Artificial Intelligence and Natural Language Processing techniques. This prototype is implemented in Python programming language and uses the social network Twitter. Moreover, we used the prototype to collect posts from Twitter and concluded that moral violence is the most common type of violence against women followed by sexual violence So, our research hypothesis was partially validated as we had assumed that most posts would refer to sexual violence. An important consequence of this conclusion is that we must also pay attention to non-physical violence against women. In conclusion, the approach proposed here provides a framework for elaboration, supervision, and promotion of public policy for the fight against gender violence. In other words, private institutions or Public Power - whether through the Executive, Legislative, Judicial, Public Ministry or Public Defenders - may use the concept or the technology proposed in this work to fight against gender violence.