NinSuna model

A tight coupling exists between the design of our Model-driven media delivery platform and a model for describing the structural, semantic, and scalability properties of media resources. This model provides support for a seamless integration of adaptation operations and content metadata. Furthermore, it allows format-independent packaging and delivery of multimedia content. 


As introduced above, the NinSuna model consists of three parts: structural, scalability, and semantic properties of media resources. A high-level overview of the model is depicted in Figure 1.

Structural model

The structural metadata part of the model describes information regarding the high-level structure of a compressed MediaBitstream. Such a bitstream consists of a list of RandomAccessUnits. Random access refers to the ability of the decoder to start decoding at a point in a compressed multimedia bitstream other than at the beginning and to recover an exact representation of the decoded bitstream. Each random access unit consists of a list of DataBlocks, pointing to particular byte ranges of the compressed multimedia bitstream. Support for existing coding formats is provided by mapping them to the structural part of the model.

Scalability model

The scalability information part of the model provides information regarding properties of the media bitstream that are related to possible adaptation operations. It enables the declaration of what types of adaptations may or should be applied to the media bitstream in order to optimally fit a given context. A DataBlock can contain ScalabilityAxisInformation, indicating the relationship between ScalabilityAxes, Features of MediaBitstreams, and their corresponding DataBlocks.

Semantic model

The semantic metadata part of the model enables to create a description of the semantic information of MediaBitstreams (i.e., AnnotatedMultimedia). Such an AnnotatedMultimedia can be decomposed into MediaSegments, which is able to represent the three fragment axes defined by the Media Fragments URI 1.0 specification. Already existing (domain-speci c) ontologies can be linked to the semantic part of the model to create detailed semantic descriptions of the multimedia content.

Publishing media metadata


The NinSuna model is partly implemented in OWL in order to publish metadata related to the served media resources. In particular, the semantic part of the NinSuna model is fully aligned with the Media Annotations ontology and extended with additional NinSuna-specific properties and support for the representation of media fragments. The ontology can be found on