Cyberbullying Classification based on Social Network Analysis
Anqi Wang, Katerina Potika · 2021
With the popularity of social media platforms, such as Facebook, Twitter, and Instagram, people widely share their opinions and comments over the Internet. The extensive use of social media has also caused a lot of problems. A representative problem is Cyberbullying, which is a serious social problem, mostly among teenagers. Cyberbullying occurs when a social media user posts aggressive words or phrases to harass other users, and that can lead to negatively affects on their mental and social well-being. Additionally, it may ruin the reputation of that media. We are considering the problem of detecting posts that are aggressive. Moreover, we try to detect Cyberbullies. In this research, we study Cyberbullying as a classification problem by combining text mining techniques, and the graph structure of the social network relationships that are based on a dataset from Twitter. We enhance a given dataset to a new one that has more information for each tweet (post). We manage to improve the classification accuracy by considering the additional social network features that we get from the user’s follower list and retweet information.