Anyone who has tried to move multimedia metadata between two systems knows the problem. One tool records a "shot", another a "segment", a third a "scene", and a fourth stores the same idea as a free-text tag. MPEG-7 gave the field a rich vocabulary for describing audiovisual content, but a vocabulary is not yet a shared meaning. Several of the accepted papers at SAMT 2006 take aim squarely at that problem.

The work on defining formal semantics of MPEG-7 profiles is a good example: rather than treating the standard as a bag of tags, it asks what each description actually commits you to, so that two systems can exchange annotations without quietly disagreeing about what they mean. That is the practical face of a multimedia ontology - not an abstract hierarchy, but a contract that lets independent components cooperate.

You can see the same instinct in the papers on knowledge-assisted analysis and on standards that bridge the multimedia and knowledge domains. The ambition is modest to state and hard to deliver: descriptions that travel. For a field that wants semantics to be computable, interoperability is not a side issue. It is the foundation that the retrieval, mining and adaptation work is built on, and it runs right through the accepted papers.