Speech recognition, image processing, and language understanding technologies have successfully been applied to broadcast news corpora to automate the extraction of metadata and make use of it in building effective video news retrieval interfaces. This paper discusses how these technologies can be adapted to cultural documentaries as represented by the award-winning First Emperor of China videodisc and multimedia CD. Through automated means, efficient interfaces into documentary contents can be built dynamically based on user needs. Such interfaces enable the assemblage of large video documentary libraries from component videodisc, CD, and videotape projects, with alternate views into the material complementing the original sequences authored by the materials' producers.
展开▼
