CbPIS: Cyberbullying Profile Identification System with Users in Loop

Abstract

The internet has given us access to vast amounts of information and enabled us to communicate globally. However, posting uncensored comments on internet has raised several concerns. Cyberbullying, provoking and pestering are some of the growing issues on social media. People try to belittle others on public forums for fun. There is no suitable and scalable way to establish the authenticity and reliability of information and users. In this paper, we propose Cyberbullying Profile Identification System (CbPIS), to identify bullying profile involved in cyberbullying. In the proposed work, we first evaluate and compare existing state-of-the-art machine learning and deep learning techniques, then select the suitable technique for the system development. Moreover, CbPIS takes Users’ feedback for validation purpose and based on the feedback, if needed, CbPIS retrains itself to improve the predictions. Experimental results show that CbPIS with user feedback gives effective results.