Cyberbullying Text Classification Using Machine Learning Approaches

Abstract

Cyberbullying has become a significant issue in today’s society, particularly among adolescents and teenagers. The rise of social media platforms and online communication tools has made it easier for individuals to harass others anonymously, often without accountability. In recent years, natural language processing (NLP) techniques have been employed to detect and classify instances of cyberbullying. These methods analyze the language used in online interactions to identify patterns and indicators of bullying behavior. This study focuses on evaluating the effectiveness of NLP techniques in detecting and categorizing cyberbullying incidents. To achieve this, the research draws on various data sources, such as chat logs, social media posts, and other forms of online communication, to understand the diverse forms of cyberbullying. The ultimate goal is to enhance our understanding of cyberbullying dynamics and explore how NLP applications can help mitigate its adverse effects. The research employs supervised learning techniques, which use labeled data to train algorithms for accurate predictions and classifications. As technology advances, it has impacted both the positive and negative aspects of life, with machine learning systems becoming increasingly effective in detecting aggressive language associated with cyberbullying. This study categorizes cyberbullying into seven groups: “Not abusive,” “gender,” “ethnicity,” “political,” “insult,” “age,” and “religion.” Among the machine learning classifiers tested, the Support Vector Machine (SVM) achieved the highest accuracy of 91.07% in identifying abusive or cyberbullying-related texts.