Document Type : Research Paper

Authors

• T. Charan Singh
• S. Srithar

Department of Computer Science and Engineering, Koneru Lakshmaiah Education Foundation, Deemed to be University, Vaddeswaram, Andhra Pradesh, 522302, India.

Abstract

Hate speech and hateful language have become more accessible to spread as a result of the increase in social media and digital contacts. Cyberbullying is the term used to describe these kinds of online insults, attacks, and harassment. Regretfully, the prevalence of cyberbullying has increased, with those who engage in it hiding behind an illusion of relative online anonymity. Finding such offensive content has become difficult due to the overwhelming amount of user-generated content. Text categorization is a broad field of machine learning. Because deep learning techniques outperform typical machine learning algorithms in various ways, researchers are turning to them to detect cyberbullying. This research proposes a new deep learning (DL)--based technique to overcome the issues of cyberbullying content recognition. To detect and classify the bullying content from pre-processed data using selected essential features, the Deep Kronecker Neural Network (DKNN) technique was employed. Comparing different classification strategies with the proposed approach, the extensive tests conducted on the two datasets demonstrate the significance of this work. We provide a novel technique for cyberbullying detection: the DKNN technique outperforms existing state-of-the-art methods with up to 99.56% accuracy results.

Keywords

• Cyberbullying
• deep learning
• detection
• algorithms
• Twitter
• cybercrime
• social media
• sentiment analysis
• cyberbullying natural language processing

Main Subjects

• Computer Science