Other Free Encyclopedias » Online Encyclopedia » Encyclopedia - Featured Articles » Contributed Topics from A-E

Context and Current Metadata Standards

descriptions content semantic mpeg

Definition: Standards that define human readable and machine understandable metadata of media content.

Maintaining human readable and machine understandable metadata descriptions of media content is essential for retrieving, using, and managing non-textual media items. Such descriptions usually contain low-level features (e.g. color histograms) extracted automatically from the underlying media object as well as high-level semantic concepts, annotated by human users. Since semantic content descriptions are always bound to human interpretation, which itself is time- and context-bound, metadata might be misattributed and not very helpful for other users in another context. Assuming that we want to facilitate the user’s ability to formulate contextual queries, it is necessary to maintain context-sensitive metadata. One of the key issues for realizing that goal is the use of context-aware data models and metadata standards. Currently there are two promising, standardized metadata frameworks that address the need for machine-processable and context-aware content descriptions: the Multimedia Content Description Interface (MPEG-7) developed by ISO/IEC and the standards developed within the W3C Semantic Web Activity.

The MPEG-7 framework is an attempt to standardize descriptions of audio-visual content by providing Description Schemes (DS) to describe different aspects of multimedia at different levels of abstraction. In MPEG-7 contextual aspects of multimedia are modeled in terms of narrative worlds. A narrative world consists of objects, agents, events, places, time, states, concepts and the relationships among those entities. MPEG-7 descriptions can contain multiple (interrelated) narrative worlds; thus content can be characterized in several contexts. The MPEG-7 Semantic DS is defined to embed descriptions of narrative worlds into standardized metadata descriptions. This approach to model context is rather closed, because the syntax and semantics of the Semantics DS is defined only within the framework and is bound to the domain of audio-visual data.

Another attempt to model context in metadata descriptions is the framework provided by the Semantic Web. The Resource Description Framework (RDF) allows for encoding complex metadata descriptions as graphs without associating any specific semantics. On top of the RDF data model, which is a directed label graph, ontology languages like RDFS and OWL were defined. Those languages have a well-defined, logic-based semantics and allow for the definition of arbitrary domain- and context-specific ontologies. In contrast to MPEG-7, which models context within a hard-wired scheme, the Semantic Web is open for any contextual model as long as it is expressed in a semantic language. Furthermore, the logical foundations of those languages are the basic requirement for reasoning services. Such services could infer additional semantics in order to add contextual metadata automatically. A first attempt into this direction is presented in – a camera phone image system that captures the media’s context. It relies on reasoning services to infer additional media semantics in order to add contextual metadata automatically.

Context-Aware Modeling of Multimedia Content [next] [back] Content Repurposing for Small Devices - INTRODUCTION, BACKGROUND, METHODS OF CONTENT REPURPOSING, Repurposing Strategies, Redrawing the Display, FUTURE WORK, CONCLUSION

User Comments

Your email address will be altered so spam harvesting bots can't read it easily.
Hide my email completely instead?

Cancel or