Effective Automatic Cyberbullying Detection Using a Hybrid Approach SVM And NLP

Cyberbullying is a growing concern in today's digital age, with profound implications for individuals' well-being and mental health. This paper presents an innovative approach to automatically detect cyberbullying in online text using a combination of Natural Language Processing (NLP), Support Vector Machine (SVM) classifiers, Term-Frequency times Inverse Document Frequency (TF-IDF), and the Linguistic Inquiry and Word Count (LIWC2) tool.The proposed system leverages the power of NLP to preprocess and analyze textual data, allowing for the extraction of essential features indicative of cyberbullying. SVM classifiers are then employed to classify text instances into cyberbullying or non-cyberbullying categories, enhancing the model's predictive accuracy. To capture the semantic and contextual aspects of the text, TF-IDF is utilized to weigh the importance of words in the document corpus. This helps in differentiating between common language and words specific to cyberbullying instances. Additionally, LIWC2 is employed to extract linguistic and psychological insights, aiding in the identification of emotional and psychological patterns associated with cyberbullying. The experiments conducted on real-world datasets demonstrate the system's ability to accurately identify instances of cyberbullying, providing valuable insights for researchers, policymakers, and online platforms in the ongoing battle against online harassment and bullying. This research represents a significant step forward in the development of automated tools to combat cyberbullying and protect online users' mental and emotional well-being. After tuning the model giving the best results, we achieve 93.15% accuracy upon evaluating it on test data. We also create a module which serves as an intermediate between user and Twitter.