| 要約: | Online social networks (OSNs) have become massively popular. The characteristics of OSNs communication is a revolutionary trend exploiting the expanded capabilities of Web 2.0, which provide users with the flexibility and freedom to post, write, and construct large social network relations. On one hand, OSNs provide users with novel and large-scale social interactions, which is a concept previously considered impossible in terms of scale and extent. On the other hand, OSNs can be used by criminals as a platform to commit cybercrimes without physically facing their victims. OSNs serve as a medium to commit cybercrimes as well as a delivery mechanism. To tackle these emerging problems, this work proposes effective methods to detect cyberbullying and identify influential spreaders in OSNs. First, an effective method to detect cyberbullying is proposed by offering a unique set of significant features, which show improvement in the performance of machine learning classifiers when compared to baseline features. Although any user in such massively connected networks can be vulnerable to online misbehavior, hence applying detection methods for every node (user) of a network is impractical. Therefore, an effective controlling method is required along with the detection method. The information spreading controlling method is achieved by proposing an effective method to identify influential spreaders in OSNs. Identifying these users is significant to either hinder the diffusion of unwanted information, such as rumor and cyberbullying, or accelerate spreading and distribution of precautionary messages as part of cyberbullying prevention strategies. Thus, interaction weighted k-core method (
|
|---|
