Published Paper Details:

The growth of social media platforms has led to a significant increase in user-generated content. These platforms have given users the ability to produce, share, and exchange content in order to interact and communicate with one another. Threatening and abusive language spreads quickly on social media, but it can be stopped if we can find and take it down. These would have, given online bullying and haters new ways to express their hate to a bigger audience while frequently remaining anonymous. We require a strong and efficient automatic system to identify threat and abusive languages due to the increased popularity of social media platforms like Facebook, Twitter, Instagram, etc. and the excessive number of social media users. Many automated methods using machine learning, deep learning, and natural language processing (NLP) have been developed in the past due to the severe and frequent nature of this activity. This paper provides a thorough summary of the reducing techniques that the research in this field has suggested for detecting offensive content. A classification of several methodologies and features used by researchers in the detection process is offered based on algorithms and methods used in detection techniques. This study also covers the major problems that require detailed research in this field. Finally, some future research directions for creating reliable social media content detection systems are also mentioned.