Friday, October 31, 2008

Workshop on the Many Faces of Multimedia Semantics

This workshop was held together with ACM Multimedia and comprised an interesting keynote about semantic video annotation using ontologies. It seems that classical knowledge engineering is approaching the search/metadata/annotation community. Anyway, the talk was very interesting at it shows a way to generate more complex rules (SWRL) from basic rules based on the ontology. That is, annotating multimedia content by applying reasoning techniques on top of ontologies.

During this workshop I learned - among others - about Multimedia OWL (MOWL) that describes spatio-temporal relationships of objects within media including a probabilistic association thereof.

Btw. I also presented our results of the MPEG core experiement related to the semantics of MPEG-21 Digital Items. People agreed that we need an equivalent of an audio/video decoder but for metadata. Current metadata specifications define the syntax (+semantics) only but no means how to "decode" metadata.

ACM Multimedia 2009 will be held in Beijing and 2010 in Florence.

No comments: