VisualCOMET+: Visual Commonsense Generation & its incorporation into a Multimodal Topic Modeling algorithm