I shared my thoughts on Multimedia Semantic Web at the International Semantic Web Conference in Karlsruhe. See the conference information here.
This was the first Keynote Talk of the conference, and was attended by more than 600 people. You can get the presentation from my homepage (www.ics.uci.edu/~jain) under presentations. The theme of the talk was bridging the semantic gap in multimedia systems. I briefly discussed EMME system as an example of our approach. I was very pleased by the interest that it generated. I think the idea that when one goes from isolated events to eventweb, some really remarkable things are possible is compelling and people easily see that. Many people approached me that they would like to use the system and collaborate in different applications.
Semantic gap has attracted lots of attention and people talk a lot about it. Much of the work in semantics is however refining and strengthening techniques strongly on both sides of the gap rather than trying to bridge the gap. There is much emphasis on ‘description’ rather than models. Descriptions are good, but for bridging the gap, one requires strong and extensive models. Description can help, but models are the bridge.
During conversations with several leading researchers in ‘Ontology’ related areas, it became clear that there is a strong group of researchers that are strong supporter of keeping ontologies close to description logic. It was clear, however, that enough number of people look at Ontologies as a mechanism to represent explicit knowledge and use this to solve real problems. They do not consider logical consistency to be the essential requirement.