The most distinctive aspect of our approach involves our initial focus on collections of spatially-indexed information. We wish to emphasize that in proposing to develop a digital library specifically for spatial data, our long-term goal is to remove the distinction between mainstream libraries, whose focus is primarily text, and special libraries of conventional materials. We believe that the unique technical problems of storing, accessing and retrieving spatial data, particularly with the use of content-based search, require us to focus initially on problems related to spatially-indexed collections.
While spatially-indexed information has been traditionally defined in a narrow way in terms of collections of facts tied to specific locations on the surface of the Earth, such as maps, atlases and images often stored in analog form, the Alexandria Project will take the much broader viewpoint that spatially-indexed data refers to data that is indexed relative to a coordinate system that is laid out with respect to some real world entity (such as the world, an human body or a painting); and/or a coordinate system that is laid out with respect to some representational medium (such as a grid on a sheet of paper or cells in a rectangular array). Based upon this viewpoint, we assume that digital libraries will ultimately require computational support for all classes of material that may be collectively defined in terms of such coordinate system indexing.
Hence, our long-term goal is to support not only library functions involving digitized maps and images, but also to support library functions in areas that involve the use, for example, of photographs, medical images, digital representations of art, architectural plans, and book and journal illustrations. In the latter case, one could envisage a request for a book in which an illustration with a certain content might be found. While the basic strategy of the project involves support for collections from the whole spectrum of spatially-indexed information, we will initially focus on collections that are indexed with respect to earth coordinate systems.
It is clear that in building a digital library in this area of application, we will be able to resolve many sets of issues that currently arise in the context of analog digital libraries. In traditional analog libraries, for example, maps, atlases, and images require special forms of cataloging and indexing; their physical dimensions require special storage; they often deteriorate rapidly with use, and convenient access by users poses major difficulties. In a digital library this issues disappear.
The Alexandria Project will also address fundamental research issues in the development of a true digital representation for spatially-indexed data. For example, we believe the conventional distinction between catalogue and materials, or between metadata and data, to be artificial particularly in the context of spatial information, and propose to allow the user to access information over a continuum of levels of abstraction. Furthermore, we will avoid incorporating the conventional and costly distinction between images, or regular rectangular arrays of values (including scanned images, remotely sensed scenes, and digital elevation models), and spatially-indexed representations of interpreted geographic features (including as air navigation charts and city street maps, with associated attributes such as text, sound, or image objects.) While the two types present different problems for content- based search, we are proposing a system design that will allow a ``seamless'' combination of images and interpreted features.