Detecting the Presence of Cyberbullying Using Computer Software
Bayzick, Jennifer and Kontostathis, April and Edwards, Lynne (2011) Detecting the Presence of Cyberbullying Using Computer Software. pp. 1-2. In: Proceedings of the ACM WebSci'11, June 14-17 2011, Koblenz, Germany.
Cyberbullying is willful and repeated harm inflicted through the medium of electronic text. Computer software was developed to detect the presence of cyberbullying in online chat conversations. Rules based on a dictionary of key words are used to classify a window of posts. A truth set of MySpace threads was created. The software was found to correctly identify windows containing cyberbullying 85.30% of the time, and it identifies innocent windows 51.91% of the time. The overall accuracy is 58.63%. This suggests that our coding rules must be refined to not falsely flag so much innocent conversation.
Repository Staff Only: item control page