Using Knowledge Representation Languages for Video Annotation and Retrieval

Effective usage of multimedia digital libraries has to deal with the problem of building efficient content annotation and retrieval tools. In particular in video domain, different techniques for manual and automatic annotation and retrieval have been proposed. Despite the existence of well-defined and extensive standards for video content description, such as MPEG-7, these languages are not explicitly designed for automatic annotation and retrieval purpose. Usage of linguistic ontologies for video annotation and retrieval is a common practice to classify video elements by establishing relationships between video contents and linguistic terms that specify domain concepts at different abstraction levels. The main issue related to the use of description languages such as MPEG-7 or linguistic ontologies is due to the fact that linguistic terms are appropriate to distinguish event and object categories but they are inadequate when they must describe specific or complex patterns of events or video entities. In this paper we propose the usage of knowledge representation languages to define ontologies enriched with visual information that can be used effectively for video annotation and retrieval. Difference between content description languages and knowledge representation languages are shown, the advantages of using enriched ontologies both for the annotation and the retrieval process are presented in terms of enhanced user experience in browsing and querying video digital libraries.