“…However, even with the apparent limitations of concept-based image retrieval, such as the cost (Layne, 1994) and the variable precision of manual annotations, contentbased image retrieval systems have failed thus far to bridge the semantic gap between indexer and end user (Datta, Li, & Wang, 2005;Enser, 2008;Lew, Sebe, Djeraba, & Jain, 2006;Neumann & Gegenfurtner, 2006;Tjondronegoro et al, 2009). Because words are the only means for describing the semantic content of images, at least for the foreseeable future, image retrieval systems will depend on text annotations, especially for image collections on social networking and photo-sharing services such as Flickr where users add textual tags and descriptions to images.…”