Harmoniurn models for semantic video representation and classificationJun YangYan Liuet al.2007SDM 2007