Determination of toxic comments and unintended model bias minimization using Deep learning approach.