Detection and Classification of Toxic Content for Social Media Platforms

Over the years the use of the internet has been increased exponentially and it is changing all the time. Recently there are two important evolutions; they are the social media platforms and mobile technology. Through this, the way people communicate with each other has changed significantly [1]. Many social media platforms like Facebook, Instagram, and Twitter, etc. have grown into worldwide networks. Massive amount of information and content arise from these social media platforms as people are engaging themselves to communicate, express their opinions and share their views daily. Even though, this type of virtual communication is highly productive and constructive, a significant part of them will be destructive. That is it is toxic in nature and some of them might include violence, obscene, threat, insult, etc. The toxic content can be either in Image or Text form. So it is essential to recognize the threat and respond to it which makes the online space more healthy and valuable. To address this problem we are proposing a system that makes use of Random Forest Machine learning algorithm and Convolutional neural network to detect toxicity in Text and Image respectively. We have successfully detected and classified toxicity in Text and Image data.