A Video Metadata Model supporting Personalization & Recommendation in Video-based Services

In this paper, we propose an MPEG-7 based video metadata model that supports personalization and recommendation in video-based services. This is achieved through the provision of adequate video retrieval support that includes video retrieval based on video content, video structure and/or video attributes, as well as video retrieval based on the relationships within videos, among videos and between videos and real world objects. Our video metadata model is structured in two layers: In the first layer, we define a set of core classes appropriate for the description of any video type. In the second layer, we define additional sets of classes, one for the detailed description of each video type. We focus here on the detailed description of the first layer of our model, while we stress, through a set of examples, the necessity of the second layer.