Target Oriented Investigation of Online Abusive Attack

The increased use of online platforms led to abusive attacks that specifically use gender- and ethnic-based and other viable characteristics to target their victims. Content abuse harms victims psychologically and socially; therefore, proper detection techniques must work together with online abuse moderation systems. The research analyzes abusive and non-abusive Twitter content, which it categorizes based on its targeted characteristics. The identification between abusive and non-abusive content processing uses MNB in combination with SVM plus logistic regression with decision tree classifier and voting classifier. The task is to identify which model offers the best accuracy in detecting abusive content without producing many incorrect positive or negative assessments. Research data collection helps experts to discover better approaches to enhance abusive text detection systems and understand the methods used for online harassment.