A cross-modal approach for extracting semantic relationships of concepts from an image database | IEEE Conference Publication | IEEE Xplore