In recent years there have been substantial improvements in the speed, power, and functionality of computer hardware and database management software. Similarly, there has been rapid growth in an ever widening variety of on-line tools to support data analysis (e.g., statistical packages, expert system shells, specialized programming languages, data visualization systems, etc.). Despite these developments, the power and sophistication of existing technology is not adequate to deal with the increasing flood of digital data demanding human interpretation and analysis (NASA, for example, expects its Earth Observation Satellites later in this decade to be sending back upwards of ten terabytes of scientific data per day).

This paper describes the design and development of a deduction-based data understanding system: IDEA (Intelligent Data Exploration and Analysis) that combines deductive, data search, and user-interface capabilities to achieve improved support far interactive data analysis. It consists of three major components: (1) a logic­ based knowledge manager that efficiently derives a proof graph comprising ali possible skeletal proofs from a users question, (2) a series of tightly-coupled commercial database management systems (currently the ORACLE, INGRES, and NUCLEUS relational DBMSs) to support data search, and (3) a user-friendly knowledge visualizer that employs a visual deductive formalism: grnnhic logic to support ease of use by the analyst.