@InProceedings{Kosch2002d, author = {Kosch, Harald and Döller, Mario}, booktitle = {Proceedings of the 10th International Conference on Multimedia, Antibes, France, Nov. to Dec. 2002.}, title = {Demonstration of an MPEG-7 Multimedia Data Cartridge.}, year = {2002}, address = {New York, NY, USA}, editor = {Rowe, Lawrence and Merialdo, Bernard and Muehlhaeuser, Max and Ross, Keith and Dimitrova, Nevenka}, month = dec, pages = {85-86}, publisher = {ACM Press}, abstract = {Multimedia Database Systems (MMDBMS) organize and store multimedia data for content retrieval. The relying multimedia data models represent abstractions of media objects for querying, indexing, and so on. However, most currently available implementations reval shortcomings. Either they are limited by one kind of multimedia data supported or by the capacity of their semantic modelling. Despite the upcoming MPEG-7 standard for representing low-level and high-level features of multimedia and respective annotation and use tools (see http://www.mpeg-industry.com), we are not aware of any MMDBMS product which integrates this standard for the purpose of a more meaningful indexing and querying. In this context, our paper presents a Multimedia Data Cartridge (MDC) for demonstration at ACM Multimedia 2002 that implements an object-relational data model for the core part of the MPEG-7 standard. It is an open and extensible system realized with the Oracle Data Cartridge technology. Oracle offers with that technology a mechanism for extending the capabilities of an Oracle database (e.g., type system, query processing and indexing) for the users needs. Besides the more effective modelling of multimedia content, efficient retrieval was considered. Efficient query processing is guaranteed through new database indexing mechanisms. Innately, most database systems provide only a limited number of integrated access methods such as B-trees. Available multimedia database extension packages (e.g., DataBlades of Informix) rarely handle indexing of d-dimensional data (e.g., feature vectors with d>2) or advanced similarity search functionalities (e.g., k-NN search). These circumstances limit the use of database systems in multimedia. For overcoming this drawback, we introduce an Multimedia Indexing Framework (MIF) relying on the GiST framework. Finally, in order to use the MMDBMS we provide query and presentation interfaces which automatically adapt to their usage environment.}, doi = {10.1145/641007.641024}, language = {EN}, pdf = {https://www.itec.aau.at/bib/files/p85-doller.pdf}, talktype = {none} }