Domain Knowledge Representation in Semantic MPEG-7 Descriptions

Some important developments in the area of multimedia applications lead to an open, Internet based environment, for multimedia. These developments include: (a) The popularity of the audiovisual content and its associated services (including the traditional TV and video services); (b) The advent of the digital media; (c) The availability of low-cost audiovisual content management devices; and (d) The development of advanced network infrastructures. In such an open environment, the syntactic interoperability provided by the standards (such as the MPEG standards) is necessary for the services offered by different vendors to interoperate. However, since the amount of the available audiovisual content grows exponentially, efficient semantic-based retrieval services should be offered, in order to allow the users to effectively manage the audiovisual content. Such services can be built on top of the semantic-based MPEG-7 audiovisual content descriptions. The MPEG-7 standard [7], which is the dominant standard for audiovisual content description, provides interoperability at the syntactic level and, at the same time, allows for the semantic description of the audiovisual content. There is, though, a serious limitation of the MPEG-7 standard: The MPEG-7 constructs intended for the semantic description of the audiovisual content are general-purpose constructs and the standard does not describe a formal mechanism for the systematic integration of domain knowledge in the MPEG-7 descriptions. Thus, the utilization of the MPEG-7 semantic description constructs, even in conjunction with textual or keyword-based descriptions of the audiovisual content, has serious limitations [5]. Consider, as an example, a query asking for the audiovisual content containing the goals of a soccer game. This approach would return, in addition to the requested material, audiovisual content that contains, in its description, the word " goal " (e.g. " shot-on-goal " , " near-goal " etc.) while it does not contain goal. It is well accepted today that the utilization of domain knowledge can improve the functionality and effectiveness of the information system applications. It can, for example, allow for reasoning on top of the content metadata descriptions and improve the efficiency of the content retrieval and filtering. The integration of domain knowledge in the metadata descriptions allows more precise querying on a semantic vocabulary which is well understood by the domain communities [2][6]. A straightforward solution for the systematic representation of domain knowledge in the MPEG-7 framework is the definition of XML Schema types that extend the general-purpose MPEG-7 types in order to represent domain-specific entities (e.g. goals …