There are quite a lot of image sites on the web. Some are socially driven and others are for stock photos. It would help if such image sites provided more semantics into their images. Images really need to be readable by crawlers. They also need to be easy to interpret as part of content extraction. Semantic ontology for images would help define at a metadata level what it is contained in them in order to provide to some degree the subliminal cues to make retrofit connections of the rich document content. It would also allow for better linked data searching. With so many dispersed stock image sites it would be useful if they all worked together in a connected way, sharing in their content and making it accessible to all. When one looks at images on the web, one notices so much that is disconnected from knowledge discovery on the web. Text and images provide for rich sources of content. There is an awful lot of content on the internet yet to be tapped by crawlers in order to penetrate the deep web and allow for more content analytics as well as ambient intelligence.
