Searching with semantics: An interactive visualization technique for exploring an annotated image collection

Abstract

Effective opportunistic search in visual interfaces for large information collections, such as documents, web pages, and annotated images, often requires domain knowledge that the information seeker does not have. Relevant semantic information, such as metadata and especially ontologies, are valuable resources for helping the user learn this domain knowledge. In this paper we describe an interactive "focus + context" visualization technique for integrating this semantic information into the representation of the information collection to support effective opportunistic search. We have implemented a prototype using these techniques that integrates WordNet, a general lexical ontology, and a large collection of professionally annotated images. As the user explores the collection, the prototype dynamically changes the visual emphasis and detail of the images and keywords to reflect the relevant semantic relationships. This interaction allows the user to rapidly learn and use the domain knowledge required for effective opportunistic search. © Springer-Verlag Berlin Heidelberg 2003.