Empirical Analysis of Supervised Machine Learning Techniques for Cyberbullying Detection

Cyberbullying is utilization of digital technology for targeting a person or a group in order to bully them socially and psychologically. Real-time social media platforms such as Instagram, Twitter, and YouTube have a large viewership, which serves as a fertile medium for such bullying activities. Instances of such harassment or intimidation are maximally found in the comments of a dynamic and an expressive medium like YouTube. This necessitates an adequate requisite to take relevant steps to find solutions for the detection and prevention of cyberbullying. The work presented in this paper focuses on the implementation of four supervised machine learning methodologies, namely Random Forest, k-Nearest Neighbor, Sequential Machine Optimization, and Naive Bayes in order to identify and detect the presence or absence of cyberbullying in YouTube video comments. The experimentation was carried out expending the Weka toolkit and utilizing the data gathered from comments obtained from YouTube videos involving core sensitive topics like race, culture, gender, sexuality, and physical attributes. The results are analyzed based on the measures like precision, accuracy, recall, and F-score and amongst the four techniques implemented, k-Nearest Neighbor is able to recognize the true positives with highest accuracy of around 83%. We also discuss various future research prospects for detection of cyberbullying.