Formalizing semantic spaces for information access

Abstract

An exponentially growing volume of digital information makes extraction of relevant items increasingly difficult. This article documents the adoption of information visualization tools by researchers in the disciplines of geography, computer science, and information science to facilitate exploration of very large data archives. Graphic depiction of database content (or the database "semantics") can be based on a spatial or even a geographic metaphor. Such depictions, often called information spaces or information worlds, provide one example of "spatialization." Various forms of spatialized views are critiqued in this article. To date, systematic approaches to the creation of spatialized views have lacked solid theoretical foundations. Three spatial frames of reference are presented to formalize and visualize semantic spatialized views: geographic space, cognitive space, and Benediktine space. Application to an example of a very large online catalog (GEOREF) highlights the underlying assumptions of the space types and demonstrates what spatial properties are preserved for each proposed approach.