Quora Based Insincere Content Classification & Detection for Social Media using Machine Learning

Internet, being a source of infinite amount of information that people use and exchange daily, is not protected from the abusive, inappropriate and toxic content. Sometimes people deteriorate the content on social media which negatively impacts the society and it may have awful results. Regarding this toxic content over social media we examined the data set of Quora, which is a question answer website, to filter the inappropriate questions which not only affects the society but also degrade the quality and the standard of such websites. This paper analysed tokenization and vectorization which are considered to be some of the best technique in natural language processing and machine learning algorithms such as Naive Bayes, Logistic Regression, Support Vector Machine and Random Forest. On the basis of accuracy, F1-score and confusion matrix we evaluated that SVM performs best result followed by Logistic Regression. The count vectorizer techniques resulted better than other text vectorizer techniques. The experimental results showed that SVM worked better by achieving an accuracy of 0.899 for the best case.