Content-driven navigation of large databases

Abstract

Advances in digital storage technology and processing speed have made feasible the creation of large databases with rapid access to individual items stored therein. While many databases still comprise textual and numeric information, an increasing number are genuinely multi-media. A desirable - and attainable - goal of such diverse databases is to retrieve information from them in a manner that is minimally constrained by the medium in which the query is started. Consider a purely imaginary example: a database of the world's endangered species. For any single entry, we might expect any or all of the following: Still images, e.g. of the species in its natural habitat; Video clips, showing aspects of behaviour; Sound recordings; Textual descriptions; Numeric data, such as statistics. If we are navigating this database, and arrive at one of these items of information, we naturally expect to be able to move smoothly to any of the others that relate to the same species. In this paper, we will concentrate on visual methods for navigating such multi-media collections but, to begin with, we briefly review some of the available strategies for searching large databases