AI Powered Cyber Bullying Detection: Enhancing Safety in Digital Communities

AI Powered Cyber Bullying Detection: Enhancing Safety in Digital Communities

Srinidhi Pudipeddi1, Shriya Aishani Rachakonda2, Dr. Alagiriswamy A A3

1 Department of Computing Technologies and SRM Institute of Science and Technology

2 Department of Computational Intelligence and SRM Institute of Science and Technology

3 Department of Physics & Nanotechnology and SRM Institute of Science and Technology

Abstract - Cyberbullying has a profound effect on its victims, and each person's reaction is unique. This variation makes it difficult to find trustworthy cyberbullying content. There is less emphasis on image-based cyberbullying identification than there is on text-based methods in certain research. The goal of this effort is to improve a model that stops hazardous image-based content from being shared on social media. We provide an automated method that recognizes dangerous cyber-symbol images from social media networks using switch learning. Images used in cyberbullying are analyzed using transfer learning to find contextual information. Two photo datasets—one with and one without photos of cyberbullying—are used in our testing. These datasets are beneficial for additional study in this field. Since finding a useful model to identify bullying photos is difficult, we investigate deep learning (DL) as well as transfer learning.

Key Words:  Cyberbullying, Data Processing, Data Collection, MoblieNet V2 Algorithm, MobilenetV2