Multi-Label Toxicity Detection: An Analysis

Everybody has the freedom and flexibility to express their opinions and ideas through social networking and online discussion platforms. However, individuals are dealing with circumstances where the vast majority takes these networks casually and misuses them as a way to abuse and threaten others, which can result in cyberattacks, cyberbullying, nightmares, and, in severe cases, suicidal attempts. Such statements must be manually identified and classified, which is a time-taking, exhausting, and unreliable process. This research examines thetoxicity of remark and also investigate the sort of toxicity, categorize the comments into various labels if they are toxic. This research work will also compare various existing machine learning algorithms on the dataset named Toxic Comment Classification Challenge imported from Kaggle that contains 15000 comments. Machine Learning, Deep Learning and Natural Language Processing lend a helping hand in reducing the poisonous environment that exists on numerous discussion forums.