Query Generalization: A Method for Interpreting Null Answers

A frustrating event in the course of interaction with a database management system is query failure: a query is submitted to the database, but instead of the anticipated printout, the system responds with an empty set of da ta items. While such null answers are always correct from a technical'point of view, quite often they are unsatisfactory. Most efforts to deal with this problem have been in the context of natural language interfaces. In this paper we outline a simple mechanism for handling query failures in a typical database management system, which has only formal language interfaces, and only limited knowledge on the data it stores (such as types and relationships). The mechanism is demonstrated with the Loose Structure data model, which adopts an object-oriented, logic-based approach. Its principles, however, may be implemented with other da ta models and user interfaces. 598 EXPERT DATABASE SYSTEMS