Detección automática de ciberbullying a través del procesamiento digital de imágenes

Abstract

In recent years, the presence in the Internet of individuals has become virtually permanent due to the growth and popularity of web technologies and social networks. Through social networks, users express feelings, opinions or emotions, where comments or discussions are then established in public or private conversations. The increase of social connections across groups of friends or followers of a user profile has generated extensive and intense debates in relation to consider carefully what is being shared and with whom. This fact is particularly relevant in the case of adolescents and young adults, since it can become an easy way for cyberstalking or cyberbullying. Thus, cyberbullying is defined as the psychological intimidation or harassment performed by an individual or several individuals against another person using information and communications technologies, which is sustained over time and committed on a regular basis and it is much more frequent among children and adolescents. It is an activity that has multiplied tremendously with the evolution and growth of social networks, where anonymity plays a key role. The purpose of this work is to perform an initial detection of cyberbullying situations through different social networking profiles, based mainly on the analysis of the images posted by users on their social network profiles. Thus, our system aims at indicating if these images have contents that may be related to a bullying activity and, consequently, it can be used as a prevention or indicator alert when a cyberbullying case appears.