MACHINE LEARNING APPROACHES FOR IDENTIFYING SOCIAL MEDIA BULLYING

Abstract

The prevalence of cyberbullying on the internet poses
significant challenges, impacting both adolescents and
adults and leading to severe consequences such as
suicide and depression. As a result, there is an urgent
need for enhanced regulation of content on social
media platforms. This study addresses these concerns
by utilizing data from cyberbullying, specifically hate
speech tweets from Twitter, to develop a model for the
detection of cyberbullying in text data. Leveraging
Natural Language Processing (NLP) and machine
learning techniques, the study analyzes a Kaggle
dataset enriched with demographic labels to construct
an effective detection model. Through comprehensive
analysis, various feature extraction methods and
classifiers, including Support Vector Machine,
Logistic Regression, Neural Networks, Random
Forest, and Naive Bayes, are examined to determine
the optimal approach. By illuminating effective
strategies for cyberbullying detection, this research
contributes to fostering a safer online environment.
Furthermore, it underscores the importance of
collaboration among stakeholders and ethical
considerations in content moderation to combat
cyberbullying effectively and protect the well-being of
internet users.