Towards a two - layered Video Metadata Model

In this paper, we propose a model for video metadata that supports video retrieval based on video content, video structure and/or video attributes. Our model supports video retrieval based on the relationships among videos and between videos and real world objects. This model is also appropriate for providing personalization and recommendation functionality in video based services, while it takes into account events covered by more than one cameras. Our model is two-layered: In the first layer, a set of core classes appropriate for supporting any video type (e.g. news, movies, football matches, etc.) is defined. In the second layer, we define a set of classes, specific for each video type, that permit a more complete description of the videos of that type. We decided to implement our model using the functionality provided by MPEG-7. 1