Report of CE on Semantic DS

The Semantic DS allows describing the world depicted by the AV content and interpreting that world, i.e., the “about” of the AV content or depicted narrative reality, which sometimes is imaginary. This document reports on the core experiment on the Semantic DS [6]. The CE originally started at the Maui meeting in December 1999 [3]. Progress reports of the CE were provided at the Geneva meeting [1] and at the Beijing meeting [2]. In Beijing, some components of the Semantic DS were promoted to the XM: Semantic DS, Object DS, PersonObject DS, Event DS, State DS, MediaOccurrence DS, SemanticTime DS, SemanticLocation DS, UsageLabel D, and some semantic relations. The main tasks of this core experiment have been the following: 1) To refine the specification of the Semantic DS by solving identified issues; 2) To define the Conceptual DSs; 3) To recommend the standardization of more semantic relations; 4) To investigate the use of membership functions to describe the strength of relations; 5) To generate simple and complex semantic descriptions of multimedia material; 6) To implement a retrieval and browsing application/s that uses the generated descriptions and that shows the functionality of the UsageLabel D, the Conceptual DSs, the State DS, and membership functions for relations, especially; 7) To recommend changes and additions to the Semantic DS based on the results of the experiment. The retrieval application that the CE has accomplished two objectives: (1) to show the utility of the components of the Semantic DS in a retrieval scenario, and (2) to be the software for the MPEG-7 XM platform.