One aspect of the Semantic Web activity is the emergence of ‘linked data’, a world-wide system of links between large-scale data sets all expressed using “lightweight” ontological techniques.
For some time, we have been investigating formal ways to express what artists say about the images they create, starting with concept maps and moving towards more formal descriptions.
These two activities come together in the design of a basic ontology for expressing properties of images and how they relate to their creation and the things they depict. In this talk we outline this ontology, with examples, and show how it can be used to fine-tune image search.
View the Warren Hayes Ai Symposium “working draft” on SlideShare, or simply the embedded presentation below.