When you ask a computer to find "a quiet beach at sunset" in a video archive, it has colours, edges, motion vectors and audio levels to work with. What it does not have, at least not directly, is the idea of quiet, or beach, or sunset. That distance between low-level descriptors and human meaning is what the community calls the Semantic Gap, and narrowing it is the organising theme of SAMT 2006.

The conference did not appear from nowhere. It grew out of two European Workshops on the Integration of Knowledge, Semantics and Digital Media Technology, EWIMT 2004 and EWIMT 2005, which between them attracted strong papers and more than a hundred participants. SAMT turns that momentum into a full conference, deliberately placing multimedia analysis, Semantic Web technologies and industrial research in the same room.

The topics give a sense of the breadth: multimedia ontologies, knowledge-assisted data mining, semantic-driven indexing and retrieval, feature fusion and dimensionality reduction, and the standards that bridge the multimedia and knowledge domains. Across three days in Athens, the programme pairs keynote talks with tutorials, workshops and an Industry Day, and closes with a session hosted in cooperation with the European Commission on the launch of the Seventh Framework Programme.

If there is a single message in the call, it is that the gap will not be closed by better pixels or better ontologies alone, but by getting the two to work together. That is the conversation SAMT 2006 is trying to host.