Modeling the Detection of Textual Cyberbullying