Abstract
The Internet has become an inevitable form of communication, which enables connections with colleagues, friends, or people with similar interests, regardless of physical barriers. However, there is also a dark side to the Internet, since an alarming number of adolescents admit they have been victims or bystanders of cyberbullying. In order to make the Internet a safer environment, it is necessary to develop novel methods and software capable of preventing and managing cyberbullying. This chapter reviews existing research in dealing with this phenomenon and discusses current and potential applications of text mining techniques for the detection of cyberbullying.
TopBackground
Bullying is not a new phenomenon, since it has existed since ancient times. However, it has acquired a novel dimension with the rise of a new environment – the Internet, and has developed a new form known as cyberbullying. Cyberbullying is a unique phenomenon associated with the use of electronic communication technologies, representing an instrument for threatening, embarrassing or socially excluding another person (Hinduja & Patchin, 2008). Bullying in the cyber environment is much crueller and more dangerous than “traditional” forms of bullying which take place in the real world. The reasons for that are primarily certain aspects of the web: persistence, the ability to search and copy, and invisible audiences (Boyd, 2007). Because of web persistence the victim is unable to hide anywhere, since the audience is not confined to the room, school yard or street, but a large online community.
Two basic characteristics of cyberspace are dominant for cyberbullying: anonymity in cyberspace and better control of social interaction in the cyber world (Dempsey, Sulkowski, Dempsey & Storch, 2011). Creating a new identity online is very easy, and can be done in a few minutes without the true identity being checked. The anonymity of the bully is enabled through the Internet. Most (84%) cyber bullies know the identity of their victims, while only 30% of cyber victims can identify the perpetrators (Ybarra & Mitchell, 2004). Another peculiarity of cyberbullying is better control of social interaction in the cyber world. Abusers can choose when they want to harass their victim, how (through which medium), and whether they wish to bully in front of an audience.
In addition, along with greater control of social interactions and added anonymity, some pupils who have lower levels of aggression in the physical environment may behave aggressively in cyberspace (Dempsey et al., 2011). Cyberbullying detection is exacerbated by the fact that victims do not inform their parents or officers in schools because they fear that the use of their phone (at school) or ability to use the Internet (at home) (Agatston, Kowalski & Limber, 2007) may be denied to them (Williams & Guerra, 2007). Ybarra et al. (2007) found that 64% of pupils who were victims of cyberbullying (or were cyber bullied), reported that they were also ”traditionally” bullied at school.
Dempsey et al. (2011) discovered that the majority of adolescents do not want to share the potential threats with adults, regardless of whether they are naive in relation to the risks in cyberspace or intentionally engaged in risky behaviour without supervision. Hence, it would be useful to inform parents about the features of new media and encourage them to supervise the way their child uses the Internet.
The main participants in the cyberbullying process are:
Key Terms in this Chapter
Text Classification: Aims to assign pre-defined classes to textual documents. by building classification model which is able to assign the correct class to a new document in the domain, or in the training process the model learns the logic of how to make a prediction.
Software for Parental Control: Helps parents in supervising their children over the Internet, and is capable of recording what is carried out on the Internet, forbidding some cites, and doing simple analysis based on key words, but cannot intelligently detect bullying in cyber space.
Text Mining: Refers to the process of the extraction of significant information and knowledge from unstructured text.
Cyberbullying: Is a unique phenomenon associated with is the use of the Internet, cell phones or other technologies to send or post a text or images intended to hurt or embarrass another person.
Social Networks: A social structure made of nodes and links, where nodes usually represents individuals or organizations. Nodes are connected with links.
Spamming: On social media is a type of spam that prevents normal interactions among users. Usually, spam (especially comment and forum spam) violate current context, because they pertain to completely different issues and topics. It may be very annoying for users.