Multi-ontology Based Multimedia Annotation for Domain-specific Information Retrieval

Recent advances in computer power, network bandwidth, information storage, and signal processing techniques have led to a proliferation of multimedia data. As a result, multimedia has become one of the most important information sources. To ensure effective utilization of multimedia assets by a variety of users, we propose a multi-ontology based multimedia annotation model, in which domain independent multimedia ontology is integrated with multiple domain ontologies in an effort to provide multiple, domain-specific views of multimedia content such that multimedia access can better address different users´ information needs. We have developed a multimedia ontology (MO) based on MPEG-7 multimedia content description tools, proposed a strategy to integrate multiple domain ontologies, and designed a term extraction procedure to automatically extract domain-specific ontological terms from textual resources of multimedia data. To evaluate multi-ontology based multimedia annotation, we have developed a multi-ontology based multimedia access platform with a Web-based interface. The preliminary evaluation suggests that multi-ontology based multimedia annotation can better meet different users´ information needs