New Frontiers for Intelligent Content-Based Retrieval

Abstract

New developments in intelligent infrastructure-based content retrieval systems are explored in this article. In this context, “intelligence”
is defined as a system’s capacity to reason and learn effectively, use context to its advantage, construct and maintain
models of the world or specific situations, and make use of dynamic representations of information. In order to build efficient
systems for retrieving audiovisual material at semantic levels comparable to human perception and cognition, we contend that
these components are necessary. We survey studies from cognitive psychology, AI, semiotics, and computer vision that provide
light on the nature of intelligence and how intelligent systems are built. We go on to talk about how content-based retrieval
systems might take use of new possibilities made possible by some of the key concepts in these areas. Lastly, we highlight a few
of our initiatives in these areas. A multimedia knowledge presentation system called MediaNet and some MPEG-7 description
tools that allow intelligent content-based retrieval are specifically introduced.