MPEG-7 is a promising standard for multimedia content description. Adequate means for the management of large amounts of MPEG-7 media descriptions are needed in the near future. Essentially, MPEG-7 media descriptions are XML documents following media description schemes defined with an extension of XML Schema named MPEG-7 DDL. However, XML database solutions available today are not well-suited for the management of MPEG-7 media descriptions. They typically neglect type information available with media description schemes and represent the basic contents of media descriptions as text. But storing non-textual multimedia data typically contained in media descriptions such as melody contours and object shapes textually and forcing applications to access and process such data as text is neither adequate nor efficient. In this paper, we therefore propose the Typed Document Object Model (TDOM), a data model for XML documents that can benefit from available schema definitions and represent the basic contents of a document in a typed fashion. Through these typed representations, applications can access and work with multimedia data contained in MPEG-7 media descriptions in way that is appropriate to the particular type of the data. Along with TDOM, we propose typing automata as an executable intermediary representation of media description schemes. Typing automata are not only capable of validating MPEG-7 media descriptions against their description schemes but also of inferring appropriate typed representations of their basic contents by exploiting type information. Thereby, TDOM and typing automata constitute solid foundations for an XML database solution that enables the adequate management of MPEG-7 media descriptions.

Top