Multimodal pLSA on visual features and tags | IEEE Conference Publication | IEEE Xplore