Detection of Behavior Patterns through Social Networks like Twitter, using Data Mining techniques as a method to detect Cyberbullying

Abstract

Social networks such as Twitter or Facebook have revolutionized the communication mechanism between human beings, but have also generated a negative impact due to inappropriate use, this fact is perpetuated by cybercriminals to hurt other people psychologically, these bad practices are called cyberbullying. This research focuses on the detection and analysis of cyberbullying on pages and with pejorative terms in Spanish, taking advantage of the power of classification of feelings through specialized tools. For the detection of cyberbullying, first the efficiency of classification of each tool is measured, through a set of pejorative terms commonly used to hurt other people. In the analysis stage we use data mining techniques to generate a dictionary of pejorative terms that are related to cyberbullying and thus be able to generate behavior patterns of these terms. And in this way provide better tools so that psychology specialists can optimize their work. The results show which platform is more flexible, also shows which is best suited to the search of incidences of cyberbullying on Twitter.