The $\mathcal{I}\mathcal{Q}$ system: A deductive database Information Lens for reasoning about textual information

Abstract

The $\mathcal{I}\mathcal{Q}$ system, described in this paper, addresses two current problems in information retrieval (IR) from document repositories: (i) the “glut-of-information” that threatens to overwhelm the users of these systems and (ii) the loss of informational value as the documents are moved from their organizational setting into a machine-readable format. The answer to these problems discussed in the paper, proposes to restore much of this information by creating a context-base to store the additional recovered knowledge. The technology used is the $\mathcal{L}\mathcal{D}\mathcal{L}$ deductive database: we demonstrate how to use the rule-extensions to declare levels of abstraction in the information that can be used to “zoomin” or “pan-out” as required by the user (an “Information Lens”). The system takes extensive advantage of the World-Wide Web and enlists the capabilities of JAVA for information display. We discuss some of the performance issues that were experienced and conclude with a description of future R&D ideas along the lines started here.

Work supported by the U.S. Department of Energy, Office of Information Management (EH-72), contract W-31-109-Eng-38.