Social Media are turning into danger for minors, particularly those who are utilizing it routinely. This activity can likewise prompt Cyber tormenting. The unstructured compositions which are accessible in the huge proportion of information can't simply be used for extra planning by PCs. Hence, specific preprocessing methods and estimations are needed to remove important models. One of the huge investigation issues in the field of text mining is Text Classification. The Twitter corpus is used as the arrangement and test data to develop an inclination classifier. The positive or negative thoughts of another tweet are used to recognize Cyber Bullying messages on Twitter using LDA with the Naive Bayes classifier. The result shows that our model gives a better outcome of precision, survey, and F-measure as practically 70%. Unsuspecting Bayes is the most fitting computation differentiating and various figurings like J48 and Knn. The CPU taking care of time for Naive Bayes count is moderately not actually the other two request computation. The show of the structure can be improved by adding extra features to more proportion of data

Keywords: Bayes classifier; Latent Dirichlet Allocation (LDA); Text Mining; Sentiment Analysis