A graph based audio-visual document annotation and browsing system

In this paper we present an audiovisual (AV) data annotation and exploitation system, which was developed as a feasibility demonstration of several theoretical models and algorithms. These models provide a graph based AV document annotation model, a rapid search method to exploit these annotations and the representation of connected, distributed graphs in XML. First we present the theoretical background underlining the new approach of our model to the audiovisual document annotation and exploitation. In the second part we describe our demonstration application illustrating the use cases related to the theoretical model. At the end we will present our plans to enhance as well the theoretical models as the application.