Research Days Poster: Cyberbullying Detection Utilizing Artificial Intelligence and Machine Learning
Document Type
Report
Publication Date
Spring 4-2023
Abstract
Technology is a tool that can be used to gain knowledge and for advancements in areas like medicine, machinery, and everyday tasks. It can be used to connect with friends, work from home, and to improve quality of life. But some social media users can use it to hurt others. Cyberbullying is a major issue that has been steadily growing over the past few years. Cyberbullying has also steadily increased the rates of stress, anxiety, depression, violent behavior, low self-esteem and may cause suicide. Cyberbullying is an ongoing problem for social media users, and it is urgent that a solution is presented. The role of Artificial Intelligence (AI) in Cyberbullying detection (CD) is currently vital. Millions of messages, photos and other sources that include hurtful, abusive or threatening content are being transmitted over the Internet daily. We use Python programming language and Google BERT Transformer to classify the text message into two categories: cyberbullying (cb) vs not. Our dataset consists of nearly 5000 words and phrases of both categories. The Neural Network model currently catches the damaging content aka cb category before it is sent with 93% accuracy. The model can further be tuned, its accuracy improved
Recommended Citation
Watson, Annaliese; Gardner, Aysha; and Moz Ruiz, Dahana, "Research Days Poster: Cyberbullying Detection Utilizing Artificial Intelligence and Machine Learning" (2023). Center for Cybersecurity. 30.
https://digitalcommons.kean.edu/cybersecurity/30