Online Hate Classifier For Social Media Platform
Harsh Kulkarni, Prathamesh Kulkarni, Vishal Khandate, Chetan Lohkare, Manikrao Laxmanrao Dhore · 2022 6th International Conference On Computing, Communication, Control And Automation (ICCUBEA · 2022
The surge in social media platforms accredits users to reveal their opinions widely. With this boost in social media usage, the impartation between users becomes divergent because of diversity in educational backgrounds, culture, country, etc. Its consequences may take the form of clashes among the people, contaminating online environments. This content that inculcates hate among the people is termed Hate Speech. Such speech comprises various aggressive, violent, offensive language to target a specific crowd sharing common ideas, philosophy, region, religion, culture, etc. To address this concern, we developed an application that detects whether a given content (text, image) contains hate messages or not. For that, we have used a dataset that contained a total of 24783 comments from Twitter platform with 77.43% of the comments tagged as offensive language and 17% of the comments tagged as hate language. We then experimented with various models like logistic regression, random forest classifier, etc. and we got the best f1 score of 0.97 for the KNN (K-Nearest Neighbour) algorithm, hence we have deployed the final outcome with the KNN algorithm.