Enabling more effective multimedia search queries

Ph.D. student Menzo Windhouwer has developed a method to enable semantic queries in multimedia databases. To provide access to stored multimedia objects, database systems use annotations, particular features describing the objects. When annotations are generated automatically, several difficulties occur. For example, the significance of annotations like cheerful or grim depends on the cultural context of the user. Windhouwer's feature grammar system can handle these ambiguities.

Publication date: 11-08-2003

Ph.D. student Menzo Windhouwer has developed a method to enable semantic queries in multimedia databases. To provide access to stored multimedia objects, database systems use annotations, particular features describing the objects. When annotations are generated automatically, several difficulties occur. For example, the significance of annotations like cheerful or grim depends on the cultural context of the user. Windhouwer's feature grammar system can handle these ambiguities. Like grammar in natural language it determines which annotations are meaningful in combination with each other. Furthermore, it allows easy maintenance of the annotations. Several case studies like an internet multimedia search engine prove the effectiveness of the system.

Windhouwer conducted his research at CWI's Database Architectures and Information Access theme (INS1). He defended his thesis on November 6, 2003.