An Intelligent Approach for Cyberbullying Detection and Prevention


  • Dr. Vijayakumar V., Dr. Hari Prasad D., Adolf P.


Social media serves as a virtual playground for cyberbullying. It allows offenders to remain anonymous. It is also very challenging to track and makes difficult to detect bullying incidents. Detections become too crucial because of multi-model data such as text, image and video etc. In this case, automated detection and prevention techniques are necessary. The proposed research is to detect, prevent and alert cyberbullying instance with the help of multi – model deep learning enabled chatbot. The hybrid model is developed by combination of LSTM and CNN to handle the multiple inputs such as text, image and video in cyberbully detection. The chatbot was built with rasa framework. The model trained with built-in dataset and experiments conducted on real time dataset shows accuracy of 85% in text based cyberbullying and 86% in image based cyberbullying. Intelligent system monitoring the communication and the intelligent chatbot prevents the cyberbullying i.e. if an instance occurs it gives alert notifications. The system proved that deep learning based cyberbully detection and intelligent chatbot effectively handles cyberbullying events in Social Media.